Patent Application
20240027981
The present application relates generally to industrial process control systems and automation systems of industrial process plants, and in particular, a next-generation architecture for industrial process control and automation systems.
For decades, distributed process control systems and automation systems of various enterprises (such as distributed or scalable process control and/or automation systems used in power generation, chemical, petroleum, or other industrial processes such as pharmaceutical or other types of manufacturing) have typically included one or more dedicated process controller devices communicatively coupled to each other, to at least one host or operator workstation via a process control network, and to one or more instrumentation or field devices via analog, digital, or combined analog/digital buses.
The field devices perform functions within the process or plant such as opening or closing valves, switching devices on and off, and measuring process parameters. Example field devices include valves, valve positioners, switches, and transmitters (e.g., devices including sensors for measuring temperature, pressure, or flow rate; and transmitters for transmitting the sensed temperatures, pressures, and flow rates). In many industrial processes, there may be hundreds, thousands, or even tens of thousands of field devices operating to send data to and/or receive commands from the one or more dedicated controller devices.
The process controllers, which are typically located within the plant environment (i.e., within the physical confines of plant and, in particular, in the vicinity of the field devices), receive signals indicative of process measurements made by the field devices (or other information pertaining to the field devices) and execute a controller application that runs, for example, different control modules which make process control decisions, generate control signals based on the received information, and coordinate with the control modules or blocks being implemented in smart field devices (e.g., HART®, WirelessHART®, and FOUNDATION® Fieldbus field devices).
Execution of the control modules causes the process controllers to send the control signals over the communication links or signal paths to the field devices, to thereby control the operation of at least a portion of the process plant or system (e.g., to control at least a portion of one or more industrial processes running or executing within the plant or system). For example, a first set of controller(s) and field devices may control a first portion of a process being controlled by the process plant or system, and a second set of controller(s) and field devices may control a second portion of the process.
Input/output (I/O) cards (sometimes called “I/O devices” or “I/O modules”), which are also typically located within the plant environment, generally are communicatively disposed between a controller and one or more field devices, enabling communications therebetween (e.g., by converting electrical signals into digital values and vice versa). Typically, an I/O card functions as an intermediary device between a process controller and one or more field devices that have inputs or outputs configured for the same communication protocol or protocols as those utilized by the I/O card.
The field devices, controllers, and I/O devices are generally referred to collectively as “process control devices,” and are generally located, disposed, or installed in a field environment of a process control system or plant. The network formed by one or more controllers, the field devices communicatively connected to the one or more controllers, and the intermediary devices facilitating communication between the controllers and field devices may be referred to as an “I/O network” or “I/O subsystem.”
Information from the I/O network(s) may be made available over a data highway or communication network (the “process control network”) to one or more other hardware devices, such as operator workstations, personal computers or computing devices, handheld devices, data historians, report generators, centralized databases, or other centralized administrative computing devices that are typically placed in control rooms or other locations away from the harsher field environment of the plant, e.g., in a back-end environment of the process plant.
The information communicated over the process control network enables an operator or a maintenance person to perform desired functions with respect to the process via one or more hardware devices connected to the network. These hardware devices may run applications that enable an operator or other user such as a configuration engineer or a maintenance personnel to, e.g., configure the process controller, the I/O devices and the field devices, change settings of the process control routine(s), modify the operation of the control modules within the process controllers or the smart field devices, view the current state of the process or status of particular devices within the process plant, view alarms generated by field devices and process controllers, simulate the operation of the process for the purpose of training personnel or testing the process control software, diagnose problems or hardware failures within the process plant, etc. The process control network or data highway utilized by the hardware devices, controllers, and field devices may include a wired communication path, a wireless communication path, or a combination of wired and wireless communication paths.
Generally speaking, a communication network (e.g., an I/O network in a process control environment) includes communication nodes which are the senders and recipients of data and communication links or paths connecting the communication nodes. Additionally, communication networks typically include dedicated routers (including firewalls) responsible for directing traffic between communication nodes, and, optionally, dedicated devices responsible for configuring and managing the network. Some or all of the communication nodes may be also adapted to function as routers in order to direct traffic sent between other network devices. Network devices may be inter-connected in a wired or wireless manner, and network devices may have different routing and transfer capabilities. For example, dedicated routers may be capable of high-volume transmissions while some communication nodes may be capable of sending and receiving relatively little traffic over the same period of time. Additionally, the connections between communication nodes on a network may have different throughput capabilities and different attenuation characteristics. A fiber optic cable, for example, may be capable of providing a bandwidth several orders of magnitude higher than a wireless link because of the difference in the inherent physical or fundamental limitations of the medium.
Industrial control system providers and users have, for many years, organized control systems for industrial processes around the Purdue Model for Control Hierarchy logical framework standardized by ISA (International Society of Automation) 95.01—IEC (International Electrotechnical Commission) 62264-1 (“the Purdue model”). The Purdue model is a network segmentation model for industrial control systems that helps conceptualize and organize concepts of industrial process architecture and, in particular, the security of the various network segments within an industrial process.
Much like the OSI model for network communications conceptually organizes computer communication networks into layers, the Purdue model divides an industrial process architecture into multiple levels and zones. Levels 0, 1, 2, and 3 represent, respectively, the physical process (e.g., the physical equipment being controlled—field devices and attendant physical I/O devices), basic control (e.g., the controllers, PLCs, etc. that monitor and control Level 0 equipment and safety instrumented systems), area supervisory control (e.g., operator workstations and the human machine interface (HMI), historian databases, configuration, etc., as well as supervisor and data acquisition (SCADA) functionality and other control logic that analyzes and acts on Level 1 data), and site operations (e.g., plant-wide control and monitoring, data aggregation, reporting, etc.), and are part of a manufacturing zone. Levels 4 and 5, respectively, represent business and logistics systems of the enterprise (e.g., database servers, application servers, file servers, etc.), and the enterprise or corporate network (e.g., a broader set of enterprise Information Technology (IT) systems, including connections to the public Internet), and are part of an enterprise zone. A demilitarized zone (DMZ) sits between the enterprise and manufacturing zones. The process control levels 0-2 generally require a higher level of trust in the safety and validity of messages, packets, and other communications, while manufacturing, corporate, and enterprise systems in levels 3-5 generally require a lower level of trust. For example, process plant systems, networks, and devices at security levels 0-3 may be protected against threats from enterprise networks at security levels 4-5, and/or from any external networks higher than security level 5 exploiting the enterprise networks, e.g., by using the DMZ and/or one or more firewalls.
As industrial processes and the associated control systems for those processes have become more complex, the operation technology (OT) that enables industrial control (i.e., the systems that monitor events, processes and devices, and make adjustments in industrial operations) has started to converge with the information technology (IT) (i.e., the systems used for data-centric computing and analysis) that has developed around it. Data from the OT system is now sought after for, and analyzed by, various IT systems. For instance, data from the operational levels of the plant may be used by various IT systems (e.g., at the enterprise level) to monitor plant efficiency, to create or update production schedules, product delivery schedules, and deliveries of input materials, and for myriad other uses. However, achieving desired levels of security within the Purdue model is exceedingly difficult, as the desired levels of security require significant infrastructure and correspondingly difficult configuration that can take upwards of a month or more during plant commissioning. Transporting the data between the layers of the Purdue model (e.g., sending data from layer 2 to layers 3 or 4), while retaining a modicum of security, has necessitated numerous security workarounds including data relays, data diodes, and a proliferation of firewall devices. In some implementations, in order to alleviate communications issues caused by these complex security features, system providers and/or site engineers have circumvented the Purdue model to send data directly from control devices to the cloud, which undermines the plant security features provided by the Purdue model.
Complicating things further, the OT systems are frequently older systems that are incompatible with generally accepted tenets of good security hygiene on IT networks, because the OT networks were frequently designed without IT security in mind. For instance, OT systems typically do not support modern identity and authentication/authorization protocols or practices, at least not at the level of the field devices and controllers. This often leads to a variety of data transfer practices that are incompatible with highly-secure networks. For example, there may be three or more domains between layers 2 and 4 of the process, and security policy may differ at each such domain. As a result, cross-layer connectivity can be challenging, leading to implementations in which holes are punched in firewalls to get data between layers, credentials are hardcoded into devices or applications and/or are passed in unencrypted form to maintain connectivity, and the like. Further still, integration of third-party software is difficult, error prone, and often insecure for the reasons described above, and vulnerabilities are frequently left unpatched because the necessary down-time would result in significant costs to plant operators. These shortcomings result in the process control network being, at best, complex, disorganized, and exceedingly difficult to maintain and, at worst, insecure.
More recently, some providers and users of industrial control systems have attempted to move portions of the industrial control system into multi-purpose computing resources such as the cloud, and to virtualize certain aspects of the industrial control system. Such attempts have had as their purpose the desire to capture and analyze ever-expanding quantities of data generated in industrial control systems, and to create virtualized redundancy, for example. However, adherence to the Purdue model and the associated security practices it requires have resulted in these attempts each suffering from one or more of the drawbacks elaborated upon above, while also having ancillary effects. Integrating cloud-based components into the Purdue model can drastically complicate the security issues by requiring data from lower levels of the Purdue model (e.g., OT systems) to traverse IT infrastructure that it was never intended to traverse. The resulting additional security infrastructure required, when added to the traversal of the IT infrastructure (whether on premise or off) can increase latency—especially with respect to control signals—to sometimes unacceptable levels.
Additionally, some providers of industrial control systems have attempted to decouple from purpose-built controller hardware (e.g., virtualize) the control algorithms that control the industrial processes. To that end, entire controllers have been virtualized so that they can execute on less specialized hardware (e.g., on servers or shared computing hardware), allowing multiple copies of the control algorithms to execute in parallel on different hardware such that one instance can serve as a backup to another, primary instance. If the primary instance fails or becomes unstable, control can shift to the backup instance, and the original, primary instance can be shut-down and re-instantiated on the same or different hardware. While such systems have some advantages in terms of controller redundancy because entire controllers can be instantiated multiple times on different hardware, and even moved between hardware, they continue to suffer from the limitations imposed by the Purdue model. Moreover, these systems require that all of the elements of the virtualized device (e.g., a controller) be replicated or moved together or at the same time, which limits the flexibility of these systems.
A new process plant and industrial control and/or automation system architecture enables significant amounts of both computer processing and IT infrastructure that is used to support a process plant, an industrial control facility or other automation facility (referred to herein as a compute fabric) to be implemented in a shared, in an offsite and/or in a virtualized manner, which alleviates many of the communications and security issues present in current process and industrial control systems that attempt to implement control using shared or virtualized computing resources, such as cloud based computing or system resources. In particular, the new control system (which may be used to implement control, monitoring and configuration activities in any process plant or industrial manufacturing or automation facility), does not attempt to follow the well-known, generally followed and accepted Purdue model when integrating and providing communications between plant devices (such as field devices), equipment controllers, supervisory controllers, site business operations and enterprise operations. As a result, the system architecture is able to implement control functions, communications and security measures in a manner that can effectively use communications and security features developed for general purpose computing uses outside of the process plant environment in a more effective manner when supporting control functions associated with a process plant or an industrial automation facility.
More particularly, the new control system architecture includes a compute fabric that is agnostic or indifferent to the physical location at which the compute fabric is implemented and includes one or more physical controlled devices (referred to herein as a pool of physical devices), such as valves, transmitters, I/O devices, etc., located at one or more specific plants, facilities, sites, or physical locations at which a product or process is being manufactured or implemented and further includes a transport network that enables or provides communications between the compute fabric and the pool of physical devices in a robust and secure manner.
Generally speaking, the compute fabric includes a physical layer comprising one or more computer processing and/or storage devices and an application layer that includes computer implemented software modules that may be implemented on the physical layer to perform various control, monitoring and configuration activities using the pool of physical devices. In one case, the application layer of the compute fabric may be implemented as one or more sets of configured containers or as a containerized system in which various different configured containers and types of containers perform different computer implemented functions with respect to the facility or enterprise in which the control system is being implemented. The physical layer of the compute fabric may be implemented in any desired computer processing, storage, and networking equipment, such as on one or more computer processors and computer databases in the cloud, on one or more computer processors and databases at one or more dedicated offsite locations separated from the plant(s), facilities, or site(s) at which the pool of physical devices are located, on computer equipment located at the physical plant or facility at which the pool of physical devices are located or any combination thereof. The new control system architecture also includes a networking layer that is disposed in between the physical layer and the application layer and that provides administration, management, and usage of physical layer resources and logical (e.g., software-based) resources as and when required for the application layer activities, and in particular to support the timing and other needs that are specific to and required by industrial process control and automation.
Still further, the various components of the application layer, e.g., the various configured containers making up the application layer, may be executed in any desired computing equipment associated with the physical layer in any desired and configurable manner. Thus, the configured containers of the application layer may be implemented in a redundant manner in various ones of the same or different computer processing equipment of the physical layer, may be moved between different computer processing equipment of the physical layer to provide for better computing and communications performance, may be replicated in various different processors or databases of the physical layer to provide for redundancy and/or for controlling replicated physical equipment, etc.
The pool of physical devices may include devices that perform physical functions utilized to control an industrial or automation processes implemented at various different locations, site, or facilities of an enterprise. For example, the pool of physical devices, which may include field devices such as valves, valve positioners, actuators, switches, regulators, sensors, etc., perform physical functions, control, and/or other types of functionality associated with controlling the industrial or automation process using the physical hardware that is located at one or more manufacturing facilities of an enterprise, and that interacts with the process material or products being manufactured to provide measurement of and control of the physical phenomena being controlled/implemented as part of the manufacturing or automation process. The pool of physical devices may be disposed or physically located at different physical locations or environments associated with an enterprise or may be entirely disposed or located at only a single physical location or environment.
During operation, the compute fabric may be communicatively connected with the pool of physical equipment disposed at one or more physical locations using one or more transport networks. The transport networks may use any desired communication infrastructure and communication protocols, including any desired wired and/or wireless communications equipment/protocol. Such protocols may be any of various process control protocols such as HART, WirelessHART, Foundation Fieldbus, Profibus, OPC UA, etc. protocols, and/or may be any of various general computing communication protocols. For example, the transport networks may use any IP based or packetized protocol, including protocols which utilize publications and subscriptions such as MQ Telemetry Transport (MQTT) and Advanced Message Queueing Protocol (AMQP). Still further, the transport networks may use or include any desired communication network physical layers, such as Ethernet, 802.11, advanced physical layer (APL), etc. physical layers. In this manner, the pool of physical devices may send data or information to and may receive data or information from one or more configured containers in the compute fabric in packetized form via one or more of the transport networks to enable the compute fabric to implement process control, monitoring and configuration activities with respect to the pool of physical devices. Further, in some embodiments, virtual networks such as VNets may be used to communicatively connect different remote infrastructures (e.g., which may be implemented via different cloud computing systems and/or via other suitable means) and to communicatively connect different physical locations (e.g., on-premises infrastructures) with remote infrastructures. For network security reasons, virtual networks may be routed through Virtual Private Networks (VPNs). For reliability purposes, different VNets may be used for different network providers (for example, AT&T and Verizon). A more detailed description of VPNs is provided elsewhere within this document.
The compute fabric may be implemented on a scalable hardware platform, portions of which can be physically located across one or more physical locations that may or may not be the same physical locations associated with the pool of physical devices. Thus, in some embodiments, at least a portion of the compute fabric may be implemented on a cloud computing platform, the hardware of which may be disposed remotely from physical location(s) of the field environment at which the pool of physical devices is located.
Generally speaking, the compute fabric supports the creation, execution, removal, maintenance, administration, and management of a plurality of containerized applications, containerized services, or other containerized components (e.g., configured containers). The pool of containerized components may include applications that have been configured into containers and/or services that have been configured into containers, each of which executes to provide a specific functionality and/or operation utilized by the control system to control, monitor and/or configure one or more of the pool of physical devices, to support process and/or automation control and system management, and to administer, maintain, and manage the system and its components across the lifetime of the system. Generally speaking, the containerized components provide functionalities that traditional process control and automation techniques typically implement via a plethora of systems, networks, computing devices, DMZs, firewalls, and applications operating across Levels 2-5 of the Purdue model, e.g., from the supervision, monitoring, and control of physical industrial processes and data acquisition at level 2 to enterprise-level IT functionalities that provide business directions and functionalities related to the system at Level 5. Further, the containerized components may provide even higher-level functionalities such as coordination and/or management between multiple systems of an enterprise or even coordination between multiple enterprises. Accordingly, and advantageously, rather than utilizing the cumbersome and resource-costly traditional architecture of Purdue Levels 2-5 and all of the numerous data diodes, firewalls, DMZs, etc. necessary to secure a process control or automation system in the traditional architecture, the control architecture described herein simply utilizes the set of containerized components executing in the compute fabric to perform the same or similar sets of process control and automation core functionalities and related functionalities without compromising security of the system, and in some arrangements, provide greater security than is possible with traditional architectures.
Moreover, different functionalities may be implemented by different containerized components within the compute fabric and so a single application or service can be configured into multiple different containerized components (e.g., different instances of the application or service being implemented in respective containers). In this manner, similar or related configured containers may execute in conjunction with different physical devices and may execute on different portions of the hardware platform of the compute fabric, for example, to create redundancy or hot spares, etc. Advantageously, various containerized components can be created (e.g., spun up) and/or removed as or when needed and, collectively, a group of containerized components may operate to form or provide a logical process control or automation system which may be implemented as a “virtual process control system” for controlling one or more industrial or physical processes.
Moreover, during execution, each containerized component can communicatively connect to a particular physical component or physical device or to another containerized component via a respective, packet-based connection over the transport network, so that each containerized component and each physical component can be identified within the system by a unique name or identity which can be associated with a particular address (e.g., IP address) within the transport network. To maintain a high level of communication security, containerized components and physical components may be authorized and authenticated on a per-component basis and optionally pair-wise with each other, e.g., by using keys or any other suitable authorization and authentication technique. Upon successful authorization and authentication, two endpoint components may communicate data, instructions, and other information with each other during a session that the two endpoint components establish over the one or more transport networks.
To further secure the system, the one or more transport networks may include one or more virtual private networks (VPNs) so that, for example, a specific containerized component communicatively connects with a specific physical component or with other containerized components using a point-to-point or peer-to-peer connection, such as a VPN, that is exclusive to only the specific containerized component(s) and the specific physical component(s). In this manner, the components may securely communicate data, messages, instructions, and/or other information with each other via the exclusive, point-to-point or peer-to-peer connection. The exclusive point-to-point or peer-to-peer connections may be established and/or torn down as or when needed. Still further, multi-point connections (e.g., VPNs or other suitable implementations) may be used to provide highly secured communications between a plurality of containerized and/or physical components. In some implementations, point-to-point or peer-to-peer networking connections may be utilized to further secure the system instead of or in addition to one of more point-to-point or peer-to-peer connections. Point-to-point connections, peer-to-peer connections, and VPNs of the transport networks may be implemented over one or more public and/or private networks, including private enterprise networks and/or the public Internet.
Advantageously, the architecture of the compute fabric abstracts (e.g., disconnects) higher level, business logic services, subsystems, and other software components of the application layer from specific computing platforms or hardware associated with the compute fabric, and enables the higher level software defined services, subsystems, and other software components to dynamically, automatically, and responsively direct and cause changes to the usage of the hardware and software resources of nodes and clusters of the computing platform using, for example, APIs, operating system (OS) support, and other services of the networking layer, without requiring any human intervention or direction. Thus, the management of the resources of the computing platform may be dynamically responsive to changes in configurations and to the needs of the higher level software defined services, subsystems, and other software components of the application layer.
In one example, industrial process control, automation, and other associated business logic is performed by higher level software defined services, subsystems, and other software components associated with the compute fabric, e.g., at the application layer. For example, a set of software defined application layer components may collectively form a logical process control or automation system that executes in conjunction with the physical components disposed at one or more physical locations or sites to implement an industrial process. Moreover, a set of third-party business logic services may also execute at the compute fabric application layer and these third-party services may be generated by a software development kit associated with the compute fabric, via which users may develop, generate, install, and manage third-party services at the application layer.
In another example, a controller or a control service in the compute fabric may be configured with one or more process control module services, parameters, and values associated with an industrial process plant, such as tags of inputs and outputs, reference values, and the like, thereby forming a configured or programmed controller service. The controller or control service may be functionally equivalent to a traditional, dedicated, hardware controller device as understood in the Purdue model, or the controller service may be functionally equivalent to a control routine or control module which is configured into and executed by the traditional, dedicated hardware controller device. A container may be configured with an instance of the configured controller service, thereby forming a container image or instance of the configured controller service that is executable to perform the specific, configured set of process control logic, e.g., by using the configured control module containers, tags, reference values, etc., when so configured. Multiple instances or container images of a configured controller service (or of other configured applications and services) may be instantiated and executed by the compute fabric.
Some configured containers in the compute fabric may be allocated or assigned to respective compute nodes of a compute fabric, and dynamically re-assigned to different compute nodes by a software defined compute service based on dynamically changing configurations, performance requirements, and other needs of the logical process control or automation system. In some situations, configured containers may be assigned (and re-assigned) to be executed by particular processors or particular processor cores of one or more compute nodes. Some configured containers, though, may be pinned to respective compute nodes and so are not dynamically re-assigned by the compute service due to dynamically occurring conditions. Configured containers may additionally or alternatively be pinned to other physical or logical components of the compute fabric. For example, a configured container may be pinned to another configured container, to a specific data cluster, to a particular processing resource (e.g., a particular physical processor or a particular physical processor core of a compute node), to a physical rack or portion of a physical rack serviced by a particular power supply (where the physical rack physically houses the hardware of one or more compute nodes), etc. Further, configured containers may be nested within other configured containers, which is particularly useful in configuring and organizing the logical process control or automation system.
The application layer of the compute fabric may include other types of application layer services such as operator displays and interfaces, diagnostics, analytics, safety routines, reporting, data historization, service configuration, communications with external or other systems, enterprise-level applications, etc. Moreover, a set of subsystems at the application layer of the compute fabric may provide or implement other virtual or logical process control-related subsystems of the logical process control system. For example, a historian subsystem may include a read service, a write service, and a search service, respective configured containers of which are nested in the configured historian subsystem container. In another example, a batch process control subsystem may include a unit procedure service, a recipe service, and a regulatory record generation service, which may be nested within the configured batch process control system container.
Generally, the subsystems allow control services and other services to be easily and coherently grouped and/or managed. In one case, each compute node of the compute fabric hosts a respective instance of each subsystem of the set of subsystems so that subsystem services are proximately and readily available to other application layer services that are executing on each compute node. Accordingly, changes to one or more of the subsystems may be coordinated among the corresponding instances thereof executing at each compute node. As such, the set of subsystems is highly and proximately available for any application layer service executing on a compute node and, in the event of a compute node failure, a compute node component failure, or a particular subsystem failure, the functionalities provided by the set of subsystems may be easily maintained for the logical process control system. The subsystems may include continuous process control, event driven process control, batch process control, state-based control, ladder logic control, historian, process user, alarm, licensing, event, version control, process configuration, process I/O, to name a few.
Still further, in some implementations, the compute fabric may implement digital twins of various software defined application services, the entire software defined application layer, various software defined support services, and/or the entire software defined networking layer. A digital twin of the target components/layers may execute in concert with the active target components/layers on top of the computing platform, and thereby receive run-time data from the field environment of the industrial process plant and operate accordingly, with the same logic, states, timing, etc. as the active target components/layers. However, the I/O and other types of data generated by the digital twin are prevented from being delivered to the field environment. In this manner, should the active targets/components fail, the digital twin of the field targets/components may simply be activated to seamlessly maintain run-time operations of the industrial process plant. Still further, in some implementations, the compute fabric may implement a digital twin of a physical component which may serve as a proxy for the physical component during run-time operations.
Likewise, the compute fabric may be used to support various enterprise level services and functions, such as real time monitoring of an enterprise functionality across multiple facilities associated with an enterprise, real time monitoring of one or more physical locations of an enterprise, providing or implementing plant or facility operator displays to enable operator input from any location, providing containerized services from any location, providing and instantiating control, other enterprise services, portions of or even an entire process control system from any location, moving the execution of services across different locations or sites, providing subscription or third-party services with respect to the enterprise, providing centralized upgrades for the enterprise, and providing centralized monitoring of the compute fabric associated with an enterprise, as well as others.
The following disclosure describes a new process plant and industrial control and/or automation system architecture relying on a shared compute fabric to implement control, monitoring and configuration activities in any process plant or industrial manufacturing or automation facility. The compute fabric is a high-performance computing system consisting of loosely coupled storage, resource management, security, networking and parallel processing functions linked by high bandwidth interconnects (such as 10 Gigabit Ethernet), and may include any one or more of a commercial multipurpose platform such as Microsoft's Azure Services platform; platforms owned, operated, and maintained by an enterprise or system provider and dedicated to implementation of process control at one or more enterprises; compute clusters situated on-premises and local to a process plant; etc. The shared resources of the compute fabric, which may be shared among process plants within an enterprise, or by multiple enterprises each operating one or more process plants, and the fact that the new architecture does not attempt to follow the well-known, generally followed and accepted Purdue model, allows a variety of improvements and innovations in system configuration, control, monitoring, and management.
While the architecture will be described in detail below, the following examples illustrate several scenarios implementing in a system the concepts described in this specification and highlight the advantages of such implementations. The examples should not be considered limiting in the functionality available, the personnel performing various tasks, the physical separation or location of various elements, or in any other manner. Instead, these examples are intended to introduce various system elements and aspects of the operation of the system, each of which will be described in greater detail elsewhere in this description.
A system provider provides and manages a compute fabric serving one or more enterprises, and provides a variety of tools and programs for configuring, commissioning, controlling, and monitoring one or more process plants using the compute fabric. These tools and programs include tools for configuring control modules and control strategies to control the process plant, tools for configuring operator workstations to monitor and control the process plant, tools for performing asset management (e.g., tracking calibration, maintenance, etc.), and tools for instantiating and managing the control modules that, after being configured and instantiated in the compute fabric, ultimately control process plant, among others. Each enterprise included in a plurality of enterprises accesses and makes use of the compute fabric and the available tools and programs to implement one or more enterprise-owned and/or -operated industrial processes in one or more respective process plants. Each of the process plants implements various aspects of its process control via a variety of containerized applications and services instantiated in the compute fabric. These include control algorithms, input-output, and security functions, among others, and the use of the containerized applications and services may facilitate various quality-of-service (QoS) features, including load balancing, fault tolerance, and redundancy implemented and managed by either the system provider or the enterprise, or by the enterprise with assistance from the provider.
With these tools available, a first enterprise owner implements at a first process plant, a refinery, a continuous process producing various products by refining a petroleum product. That process plant has disposed therein a variety of field devices (e.g., valves, tanks, distillation columns, sensors/transmitters, and the like) that sense parameters in the refinery and perform control actions in response to control algorithms that use as inputs the sensed parameters. Each of the field devices has a corresponding input/output device that receives the signal(s) from the field device, coverts the signals into a common format (e.g., Ethernet packets), and transmits the data from the field devices to the compute fabric. A pre-configured gateway/router/aggregator, facilitating secure communication between the first process plant (e.g., I/O devices and field devices, collectively comprising a set of physical devices) and the compute fabric, is among the only non-field device or I/O device hardware that is located on the premises of the first process plant.
A configuration engineer for the enterprise and/or the process plant accesses the tools made available by the system provider to configure the operation of the process plant. The configuration engineer creates the necessary control algorithms by instantiating function blocks for receiving data from the field devices via the I/O devices and sending commands and data to the field devices via the I/O devices, instantiating various processing function blocks that utilize the data received from the field devices as input to a control algorithm and generate the outputs that are sent to the field devices, and implementing operator workflows that allow the plant operators to monitor and respond to conditions in the run-time process plant. However, in contrast to known, conventional or traditional systems, these function blocks and control modules are not downloaded to a dedicated hardware controller at the process plant. Instead, once the process plant is commissioned (e.g., the physical devices are installed and wired at the process plant), the function blocks and control modules are instantiated as containerized services and/or other types of micro-encapsulated execution environments (“MEEES,” also referred to herein as “microservices” or “granules”) in the compute fabric.
Generally, a microservice, granule, or MEEE which is instantiated in the compute fabric may be an independent software process that can run on its own deployment schedule and can be updated independently of other microservices. Examples of MEEEs may include function blocks, control modules, control applications, and other applications and services which relate to business logic of the process plant and/or otherwise support the process plant, to name a few. Groups of microservices or MEEEs may interact collaboratively to achieve some desired outcome. For example, to control a reactor, multiple strategies such as Feed, Reactor, Products, Utilities, and Flare may defined by respective MEEEs, and this set of multiple MEEEs may cooperatively operate (e.g., in conjunction with each other) during run-time of the process plant to implement the desired reactor control strategy. In another example, for a process control analytics application, various MEEEs may be defined to execute respective statistical computations and/or statistical algorithms, and the various MEEEs may be combined as desired and executed in combination to provide an overall prediction analytics application. A single, individual microservice or MEEE may be configured to execute an application ranging from very broad (e.g., a control strategy of an entire system or plant) to very granular (e.g., only a portion of a control routine or control module), as will be discussed in more detail elsewhere within this document. As such, due to its flexibility of being configurable to execute various process control and process control-related applications ranging from broad to granular, a microservice or MEEE is interchangeably referred to herein as a “granule.”
At any rate, the configuration engineer may also use the configuration tools to specify various QoS metrics for each of the function blocks and control modules, for individual signals or variables, and for the process as a whole. Each of the microservices, MEEEs, or granules communicates with one another and with the I/O devices via one or more secured point-to-point (PTP) and/or peer-to-peer (P2P) connections (which may include one or types of secured, encrypted PTP and/or P2P connections, such as VPN(s)), and each is authenticated via a digital security certificate. These secured, point-to-point and/or peer-to-peer connections and security certificates are managed automatically within the compute fabric with minimal to no input from the personnel at the enterprise.
With respect to the QoS metrics, an orchestration service operating within the computing fabric (and provided by the system provider) implements various load balancing and redundancy schemes to ensure that the QoS metrics for the plant are met. For example, the orchestration service ensures that a redundant configured container is always instantiated for any configured container executing any portion of the control algorithm, and that the redundant configured container maintains corresponding inputs and outputs (i.e., maintains a parallel state) to the primary configured container such that, if there is a failure of the primary container, control may be shifted to the redundant container nearly instantaneously (e.g., within milliseconds). The orchestration service ensures that configured containers are executing on separate hardware and powered by separate power supplies to maintain sufficient redundancy according to policies set by the enterprise. For some microservices, configured containers, MEEEs, or granules, the orchestration service instead maintains redundant state databases storing the state of the microservices/configured containers/MEEEs/granules, such that if a microservice/configured container/MEEE/granule fails or is otherwise taken offline, a new microservice/configured container/MEEE/granule may be instantiated nearly instantaneously (e.g., within milliseconds) and restored to the state of operation of the previous instantiation of the configured container when it was taken offline.
The orchestration service also provides load balancing services to maintain sufficient memory, processing, and network resources to meet QoS requirements for individual microservices/MEEEs/granules and for the plant as a whole. For example, maximum latency requirements require certain configured containers to be executing on compute fabric resources that are physically closer to the process plant and/or have greater network bandwidth between the resources and the process plant.
To maintain security, and as described above, all containerized applications and services communicate via one or more secured, encrypted point-to-point connections, such as VPNs. In some cases, a pair of endpoints (e.g., a pair of containerized applications or services, or a containerized application/service and a physical component) communicates via a dedicated VPN, while other VPNs include multiple containerized applications/services communicating with one another via respective sessions established over the VPNs. Other VPNs facilitate communication between enterprise-level containerized applications/services and plant-level containerized applications/services, other VPNs still facilitate communication between provider level containerized applications/services and containerized applications/services at the enterprise and/or plant level, and still other VPNs facilitate communications between user interface devices and the system. In any event, any human user, and any third-party application executing to facilitate a service in the enterprise or the process plant, interacts with the system one or more APIs through which necessary actions may be performed after the user or third-party application is authenticated (e.g., using multi-factor authentication).
In this manner, relative to known systems, the control of the first process plant is implemented with fewer dedicated computing resources, while maintaining (or improving) QoS metrics, maintaining (or even improving) security, and removing or reducing the need to periodically and manually adjust the type or quantity of local computing resources according to changes in the process plant.
At some point after the commissioning of the first process plant, the first enterprise owner decides to duplicate the first process plant in a second process plant. Because the control algorithms and necessary software are already configured for use with the first process plant, setting up the field devices and I/O hardware at the second process plant is among the most time-consuming portions of commissioning the process plant.
In this case, the enterprise owner opts to remove the physical I/O devices from the process plant setup and instead perform the functions of the I/O devices as microservices, MEEEs, or granules instantiated in the compute fabric. The enterprise owner installs the field devices at the site of the second process plant. Each field device is configured as it normally would be—with a device tag, ranges, limits, scale, and other data for operating the field device. Because the second process plant is a duplicate of the first process plant, each field device is configured and programmed according to its corresponding device in the first process plant, using an automated process to decrease the time required for commissioning. Each device is coupled, by its corresponding communication media, to a media converter that converts between the device's native communication media (e.g., Foundation Fieldbus, HART, WirelessHART, OPC UA, 4-20 mA, Ethernet, etc.) and an Ethernet protocol. Each media converter packetizes (in Ethernet packets) the variety of data received from the corresponding field device and transmits the packets to a pre-configured on-site gateway of the second process plant. The gateway facilitates secure communication between the second process plant (media converters) and the compute fabric, and is among the only non-field device hardware that is located on the premises of the second process plant.
The configuration engineer, having opened the stored configuration of the first process plant within a user interface application provided by the compute fabric, drags a loop over the compute fabric canvas (the workspace showing for the first process plant the configured containers instantiated in the compute fabric) to select all of the configured containers and, by dragging the selected configured containers to a new compute fabric canvas, copies to the second process plant the set of configured containers (e.g., containerized applications and services) instantiated in the first process plant.
The configuration engineer instantiates, within the compute fabric, for each field device of the second process plant, a digital twin of the field device and corresponding I/O microservices/MEEEs/granules, taking the place of the physical I/O devices that would previously have coupled physical field devices to the controllers. This digital twin is, to the rest of the process control software executing in the compute fabric, indistinguishable from the hardware field device operating in the process plant. The digital twin is configured identically, and maintains to the extent necessary (and possible, as described below) the same data as is present on the field device itself. That is, as the field device updates the values of measured parameters or status values, those values are uploaded, via the media converters, to the digital twins within the compute fabric. It is the digital twins, however, that interact with the instantiated function blocks, control modules, and other control algorithms in the compute fabric. By the same token, the function blocks, control modules, and other control algorithms instantiated in the compute fabric (or in a hardware controller) send commands and data to the digital twins, which communicate those commands and data back to the hardware field devices of the second process plant via the media converters.
With the hardware field devices in place and coupled via the media converters to the digital twins executing in the compute fabric, the configuration engineer instructs the compute fabric to bring the process online at the second process plant, e.g., by activating a single user control. Responsively, the compute fabric initializes the process control system, and the system is online and ready to begin testing and/or operating within minutes thereafter, vastly simplifying the process of bringing the second process plant online.
In addition to simplifying its commissioning, the digital twins contribute to the robustness of the operation of the second process plant. In the event that a particular field device becomes unavailable (momentarily or otherwise), the digital twin may prevent an abnormal condition in the operation of the process plant as a whole. For example, a momentary loss of connectivity or increase in latency between the process plant and the compute fabric may have no effect on the process as a whole, because the digital twins may continue to provide data (e.g., simulation data, presumed steady-state data, etc.) to the control algorithms (within safety parameters, of course) during the momentary anomaly. Similarly, if the self-reported (or system-determined) status for a sensor changes to indicate that the sensor value is no longer reliable, the digital twin may be programmed to provide data from another source (e.g., simulated values, values calculated based on other measured parameters, etc.) in place of the unreliable sensor data.
The use of the digital twins also contributes to the cost-effective maintainability of the process plant. In the event that a hardware sensor (e.g., a thermocouple or pressure sensor) fails, the sensor can, in many instances, be replaced without stopping or disrupting the operation of the process, because the digital twin can continue to provide data (simulated, or otherwise) to the control algorithms even when the hardware sensor is offline.
With the first and second process plants up and running, the enterprise owner (or its users) can manage both of them at an enterprise level using a variety of tools available from the system provider. Because the various facilities owned by the enterprise are executing in the compute fabric, the enterprise owner may securely access data related to the process plants individually, or to the enterprise as a whole, at any time and from anywhere. As indicated above, all human and/or third-party application access to the system is provided via APIs, access to which requires multi-factor authentication. Thus, after authenticating, the user can access tools, metrics, and other functions that allow for the management of the enterprise and its various process plants.
The enterprise-level user can create or use any number of dashboards and tools for facilitating enterprise-level views of the process plants individually or collectively. The enterprise-level user may decide to view real-time production metrics (e.g., production output volumes, quality metrics, etc.) for a single plant, and also to compare real-time metrics between plants, charting metrics for the respective plants over time for comparison. Upon noticing that one plant is performing differently from the other (e.g., outputting higher quality product, operating more efficiently, etc.) the enterprise-level user may decide to dig more deeply into why the plants are performing differently. Turning to an application or service marketplace hosted by the system provider, the enterprise-level user may purchase or subscribe to an analysis service or tool, which, upon purchase or subscription is instantiated in the compute fabric and can be used by the enterprise-level user to analyze the performance of the process plants.
The analysis shows that the tuning of one of the process plants could be optimized to perform better, and recommends a different tool that can re-tune and optimize the process plant. The enterprise-level user may reach out to the control operators at the plant to inform them of the available tool. The tool is then subscribed to or purchased for the enterprise or for just the individual plant, and instantiated in the compute fabric at the plant level and/or at the enterprise level.
The tool may determine that the process plant requires retuning in a manner that requires an updated copy of one or more control algorithms. In that event, the tool proceeds (with input from the operator) to create updated control algorithms. The updated control algorithms are instantiated in the compute fabric, but are not initially placed in control of the process plant. Instead, the updated control algorithms are implemented in parallel (e.g., in simulation mode) with the active control algorithms to ensure that they will not adversely affect the operation of the process plant and will, in fact, improve the operation of the process plant. The operator then switches control of the process to the new control algorithms instantaneously (e.g., without interrupting the process) when satisfied that the updated control modules will function properly.
At the same time, the operator of one of the process plants may notice that the tuning of the process for which the operator is responsible appears to be fluctuating. The operator reaches out to a customer service representative from the system provider for assistance to determine what is causing the fluctuation in tuning, and using dashboards available to the system provider representative, they determine that the fluctuations in tuning may be the result of several related factors in the compute fabric including the movement of microservices/MEEEs/granules between hardware resources for load balancing, changes in available processing and network bandwidth, and physical distance between the compute fabric resources and the process plant in question. The customer service representative may recommend the use of a real-time tuning application.
The real-time tuning application monitors latency between the compute fabric and the physical resources in the process plant, and automatically adjusts the tuning of the control algorithms to account for changes in the latency as a result of network conditions, physical distance, and processor bandwidth. In this example, after implementing the real-time tuning application, the operator notices that the subject process is generally more stable.
However, if the real-time tuning application indicates that there is one control loop that the real-time tuning application is unable to automatically tune, the operator and customer service representative, working cooperatively, may determine that the control loop in question requires minimal latency and, as a result, the customer service representative may “pin” the configured containers related to the control loop to physical hardware resources in the compute fabric that meet the latency requirements and, in particular, that meet the latency requirements because of their physical proximity to the process plant in question. For good measure, the customer service representative may additionally dedicate specific hardware resources for the configured containers related to the control loop, to prevent those resources from becoming loaded to an extent that would cause latencies to detune the process.
The enterprise owner, now having multiple process plants running, may determine that it would be more efficient to consolidate the operators that manage the various processes. While some number of maintenance and other personnel must be physically present at each process plant, the fact that the compute fabric is securely accessible, essentially from anywhere with a sufficient network connection, allows the operators to be located anywhere. Accordingly, the enterprise owner may determine that it can consolidate all of its operations into three operations centers, spaced roughly equally around the world, such that its operations can continue 24 hours/day without any of its employees working outside of first shift (business) hours in their respective locale.
The enterprise owner staffs each operations center with operators in numbers sufficient to operate all of its plants operating around the world. As a result of the consolidated operations centers, the numbers of staff required for redundancy (e.g., to account for staff illnesses, vacations, etc.) is reduced.
Separately, a second enterprise that is looking to improve efficiency in its traditionally-architected (first) process plant and to expand to add additional process plants desires to do so by implementing a system provided by the system provider. The system provider establishes accounts for the second enterprise, and the second enterprise personnel subscribe to and/or purchase the necessary and desired tools and packages. The second enterprise's configuration engineers use tools available from the system provider to convert the configuration files that currently operate the legacy controllers in the first process plant into a configuration of containerized services (e.g., microservices, MEEEs, or granules) that will execute on the compute fabric. At the same time, the system provider arranges to install, at the site of the first process plant, a pre-configured gateway securely coupling the I/O devices to the compute fabric. After ensuring that the compute fabric is configured to meet the necessary QoS metrics, the configuration engineers and operators of the first process plant simulate operation the process plant using the configured compute fabric to confirm that it appears to be properly configured, and then bring the process plant online using the compute fabric.
While the newly reconfigured process plant is operating using the compute fabric, and the legacy controller hardware is no longer operating the process plant, the second enterprise owner, rather than decommissioning the legacy controller configuration, maintains it. In fact, the system provider assists the enterprise personnel in configuring the legacy configuration of the first process plant such that, in the event that the compute fabric (or the network connection thereto) becomes unavailable or unstable, control of the process plant can fail over to local control using the legacy system. As such, the legacy system is run in parallel as a backup control solution.
At the same time, the second enterprise owner may decide to keep the safety-instrumented systems (SIS) for the process in place at the first process plant, rather than migrating them to the compute fabric.
As the second enterprise owner moves toward expanding to add new process plants, the second enterprise owner purchases the necessary field devices for installing in and operating those process plants. Each of the field devices includes a burned-in hardware ID that indicates its make, model, options, and the like. Upon purchasing the field devices, the burned-in hardware IDs for the purchased devices are registered to the second enterprise owner in a database that facilitates the configuration of the field devices in the new process plants. The configuration engineers, while configuring the control modules, can select each device from the database, causing the compute fabric to create a digital twin for each that is programmed accordingly. A plant engineer configuring the physical field devices in the new process plant scans each device and places each device according to its hardware ID. When the physical devices are connected to the compute fabric via their respective media converters, each is automatically associated with its digital twin based on the hardware ID, and is programmed (if programmable) according the programming of its digital twin.
While the additional process plants are configured to operate on remote compute fabric, the second enterprise owner decides that, because only one network provider serves some of the process plant locations, it remains prudent to have a control solution that can maintain safety, if not full operation, of the process plant locations in the event that the network connection to the remote compute fabric becomes unavailable. For that reason, the enterprise owner physically locates certain compute fabric resources on-site such that, in the event of a communication failure between a process plant and the remote compute fabric, the on-premises compute fabric can maintain safety and/or operation of the process plant. The on-premises compute fabric resources execute redundant containerized services, microservices, MEEEs, or granules (orchestrated by the orchestrator service in the compute fabric) such that control of the process plant can fail over to the on-premises compute fabric if necessary.
The NGPCAS 100 includes a compute fabric 102 communicatively connected to a plurality (e.g., a pool) of physical devices 105, 108. The plurality or pool of physical devices 105, 108 includes devices that perform physical functions utilized to control an industrial or automation process provided by an enterprise. For example, the plurality of physical devices 105, 108 may include field devices such as valves, valve positioners, actuators, switches, regulators, sensors, (e.g., temperature, pressure, level and flow rate sensors), spectrometric devices, pumps, motors, transmitters, and the like, some of which may be intelligent field devices. Some of the physical devices 105, 108 may have respective on-board processors, memories, and computer-executable instructions stored on the memories, where the stored computer-executable instructions are executable by the on-board processors to perform, for example, control and/or other types of calculations, alarming functions, and/or other functionality associated with controlling the industrial or automation process using the physical device 105, 108. Physical devices 105, 108 may responsively operate and/or change their behavior based on control signals and/or other instructions received from the compute fabric 102, as is described in more detail elsewhere herein.
The pool of physical devices 105, 108 of the system 100 may be disposed or physically located at different physical locations, sites, plants, or environments 115, 118 (such as illustrated in
Each physical location or environment 115, 118 at which at least one physical device 105, 108 of the system 100 is disposed includes one or more on-premises physical I/O (Input/Output) interfaces 125, 128 configured to receive, condition, and deliver (e.g., to the compute fabric 102 via one or more transport networks 130) I/O signals or I/O data generated by the on-premises physical devices 105, 108, and optionally to provide control signals, instructions, and/or other information generated by the compute fabric 102 and received at the location 115, 118 via the one or more transport networks 130 to designated recipient on-premises physical devices 105, 108. Each on-premises physical device 105, 108 physically connects to the on-premises physical I/O interfaces 125, 128, e.g., via respective wired or wireless links. As such, in an embodiment, the on-premises physical I/O interfaces 125, 128 may include a pool of I/O hardware ports (which may include wired and/or wireless ports or interfaces) and/or other I/O hardware resources that are shared among multiple on-premises physical devices 105, 108 at a respective location 115, 118. Additionally or alternatively, the on-premises physical I/O interfaces 125, 128 may include individual instances of I/O hardware resources, where each individual instance is included in or exclusively connected to one and only one on-premises physical device 105, 108. Generally speaking, the present disclosure utilizes the term “physical component” 135, 138 to collectively refer to the combination of a single physical device 105, 108 (e.g., a single field device) and the physical I/O interface 125, 128 utilized by the single physical device (e.g., the attendant physical I/O interface(s) of the single physical device 105, 108) to communicate information over the transport networks 130. As such, terminology-wise, the NGPCAS 100 includes a pool of physical components 135, 138, each of which includes an individual field device 105, 108 and the respective physical I/O interface resources 125, 128 that the individual field device 105, 108 utilizes to communicate with the compute fabric 102. Generally speaking, the physical components 135, 138 of the NGPCAS 100 operate or would be included in Level 0 of the Purdue model of a traditional process control system.
As shown in
In some embodiments, a physical site or location 115 may include a gateway/router/aggregator 148 which, for ease of discussion, is referred to herein as a “gateway 148.” Generally speaking, the gateway/router/aggregator 148 receives outgoing data and information that is to be sent to the compute fabric 102 and causes the outgoing data and information to be sent over the transport networks 130 (e.g., in individual packets and/or in packets in which data/information generated by multiple physical components 138 are aggregated into a single packet), and the gateway/router/aggregator 148 receives incoming data and information that is received from the compute fabric 102 (e.g., in individual packets and/or aggregated packets), and routes the information, instructions, and/or data included therein to designated recipient physical components 138 at the site 115. A physical location may include respective gateway 148 (e.g., location 115), a physical location may exclude any gateways 148 (e.g., location 118), and in some configurations, multiple physical locations may share a single gateway 148 (not shown in
Turning now to the compute fabric 102 of the NGPCAS 100, the compute fabric 102 is implemented on a scalable hardware platform, portions of which can be physically located across one or more physical locations (not shown in
The compute fabric 102 of the NGPCAS 100 supports the creation, execution, removal, maintenance, administration, and management of a plurality or pool of containerized applications and/or services 140, that are generally referred to interchangeably herein as a “plurality or pool of containerized components 140,” a “plurality or pool of micro-encapsulated execution environments 140 (MEEEs 140),” or a “plurality or pool of granules 140” of the NGPCAS 100. That is, the pool of containerized components/micro-encapsulated execution environments/granules 140 may include applications and/or services that have been configured into containers and/or other types of micro-encapsulated execution environments or granules, each of which can execute to provide a specific functionality and/or operation utilized by the system 100 to control the physical devices 105, 108, to support process and/or automation control and system management, and to administer, maintain, and manage the system 100 and its components across the lifetime of the system 100. Generally speaking, the containerized components/MEEEs/granules 140 of the NGPCAS 100 provide functionalities that traditional process control and automation techniques typically implement via a plethora of systems, networks, computing devices, DMZs, firewalls, and applications operating across Levels 1-5 of the Purdue model, e.g., from the basic control of physical industrial equipment and processes of the system 100 at Level 1 to enterprise-level IT functionalities that provide business directions and functionalities related to the system 100 at Level 5. Further, the containerized components/MEEEs/granules 140 may provide even higher-level functionalities such as coordination and/or management between multiple systems 100 of an enterprise or even coordination between multiple enterprises. Accordingly, and advantageously, rather than utilizing the cumbersome and resource-costly traditional architecture of Purdue Levels 1-5 and all of the numerous data diodes, firewalls, DMZs, etc. necessary to secure a process control or automation system in the traditional architecture, the NGPCAS 100 simply utilizes the set of containerized components/MEEEs/granules 140 executing in the compute fabric 102 to perform the same or similar set of process control and automation core and related functionalities without compromising security of the system 100, and often, with increased security for the system 100. A more detailed discussion of security techniques utilized within the architecture of the NGPCAS 100 is provided elsewhere within this disclosure.
Typically, different functionalities are implemented by different containerized components/MEEEs/granules 140 within the compute fabric 102. A single application or service can be configured into multiple different containerized components/MEEEs/granules 140 (e.g., different instances of the application or service implemented in respective containers), if desired, for example, to execute in conjunction with different physical devices, to execute on different portions of the hardware platform of the compute fabric 102, to create redundancy or hot spares, etc. Various containerized components/MEEEs/granules 140 can be created (e.g., spun up) and/or removed as or when needed by the system 100. Collectively, a group of containerized components/MEEEs/granules 140 may operate to form or provide a logical process control or automation system 145 (also referred to herein interchangeably as a “virtual process control system” 145) for controlling one or more industrial or physical processes by controlling and utilizing the physical components 105, 108 disposed in the field environment 120 of the NGPCAS 100. Typically, but not necessarily, the group of containerized components/MEEEs/granules 140 forming the logical process control system 145 is a subset of the entirety of containerized components/MEEEs/granules 140 provided by the compute fabric 102.
During execution, each containerized component/MEEE/granule 140 can communicatively connect to a particular physical component 135/138 or to another containerized component/MEEE/granule 140 via a respective, packet-based connection over the transport networks 130. As such, each containerized component/MEEE/granule 140 and each physical component 135/138 of the NGPCAS 100 is identified within the system 100 by a unique name, identifier, or identity which can be associated with a particular address (e.g., IP address) within the transport networks 130. Generally speaking, a physical component 135/138 may be a sender or provider of I/O data and information that one or more containerized components/MEEEs/granules 140 receive or consume. In some scenarios, a containerized component/MEEE/granule 140 can be a sender or provider of control signals or other instructions that a physical component 135/138 receives or consumes. In some scenarios, a containerized component/MEEE/granule 140 can be a sender or provider of data that another containerized component/MEEE/granule 140 receives or consumes. To maintain security of the NGPCAS 100, containerized components/MEEEs/granules 140 and physical components 135/138 may be authorized and authenticated on a per-component basis and optionally pair-wise with each other, e.g., by using keys or any other suitable authorization and authentication technique. Upon successful authorization and authentication, two endpoint components 140, 135/138 may communicate data, instructions, and other information with each other during a session that the two endpoint components 140, 135/138 establish over the one or more networks 130.
To further secure the NGPCAS 100, the one or more transport networks 130 may include one or more secured, point-to-point (PTP) and/or peer-to-peer (P2P) connections, which may include one or more secured, encrypted point-to-point and/or peer-to-peer connections such as virtual private networks (VPNs) and other types of secured, encrypted PTP and/or P2P connections. In an embodiment, a specific containerized component/MEEE/granule 140 communicatively connects with a specific physical component 135/138 using a secured, encrypted point-to-point or peer-to-peer connection (such as a VPN or other suitable implementation) that is exclusive to only the specific containerized component/MEEE/granule 140 and the specific physical component 135/138. For example, the specific physical component 135/138 and the specific containerized component/MEEE/granule 140 can be the endpoints of an exclusive point-to-point VPN that is utilized by only the two endpoint components 135/138 and 140 (e.g., and by no other components of the system 100) so that the components 135/138, 140 may securely communicate data, messages, instructions, and/or other information with each other via the exclusive, point-to-point VPN. In a similar manner, a particular containerized component/MEEE/granule 140 can communicatively connect with another particular containerized component/MEEE/granule 140 using a secured, encrypted peer-to-peer (P2P) connection (which may or may not be implemented via a VPN) that is exclusive to only the two containerized components/MEEEs/granules, and the two containerized components/MEEEs/granules may securely communicate data, instructions, and/or other information with each other via their exclusive, secured and encrypted peer-to-peer connection. The exclusive point-to-point and/or peer-to-peer connections between a single containerized component/MEEE/granule 140 and either a single physical component 135/138 or a single other containerized component/MEEE/granule 140 may be established as or when needed, and may be torn down when desired (e.g., upon completion of a data exchange, when system resources need to be freed up, etc.).
In an embodiment, a single physical location 115, 118 can be communicatively connected to the compute fabric 102 via a secured, encrypted location-to-location PTP or P2P connection (such as a location-to-location VPN or other suitable implementation) that is exclusive to the specific location 115/118 and the compute fabric 102 (not shown). Physical components 135/138 disposed at the specific location 115/118 and containerized components/MEEEs/granules 140 executing on the compute fabric 102 may communicate data and information to each other via the location-to-location PTP or P2P connection. To illustrate, in the example arrangement shown in
In an embodiment, a multi-location (e.g., multi-point) secured, encrypted connection may exclusively service the compute fabric 102 and the physical components 135/138 that are disposed at multiple, selected physical locations 115, 118 of the system 100, where the selected physical locations are a subset of an entirety of all physical locations serviced by the system 100. In this embodiment, each of the multiple physical locations 115, 118 may include a respective on-premises gateway 148 to the multi-location secured, encrypted connection, and may be associated with one or more gateway applications 150 executing on the computing fabric 102. Various physical components 135/138 disposed at the multiple locations and the containerized components/MEEEs/granules 140 corresponding to the multiple locations may authenticate to the multi-location secured and encrypted connection, and may establish sessions over the multi-location secured and encrypted connection (e.g., via the one or more gateway applications 150 and the respective on-premises gateways 148) for sending and receiving data and information to/from other components 140, 135, 138. A multi-location, secured encrypted connection may be implemented by using a VPN or other suitable mechanism, as desired, such as multiple PTP and/or P2P connections, point-to-multipoint (PTM) connections, etc.
In an embodiment, all of the containerized components/MEEEs/granules 140 of the compute fabric 102 and all of the physical components 135/138 of all of the physical locations 115/118 of the NGPCAS 100 may be serviced by a single, system-wide secured, encrypted connection, which may be implemented as a system-wide VPN or other suitable type system-wide, secured, encrypted connection. Each of the containerized components/MEEEs/granules 140 and the physical components 135/138 of the system 100 can authenticate to the system-wide connection, and can establish sessions with other components 140, 135, 138 over the secured, encrypted system-wide connection (e.g., via respective on-premises gateways 148 and one or more gateway applications 150) for sending and receiving data and information to/from the other components 140, 135, 138.
Further, if desired, various secured, encrypted PTP, P2P, PTM, and/or multi-point connection techniques may be combined (e.g., by utilizing subnets) to provide even more security. For example, a specific physical component 135/138 and a specific containerized component/MEEE/granule 140 can establish and utilize an exclusive point-to-point VPN as a subnet of a location-to-location VPN, a group of containerized components/MEEEs/granules 140 may be included in a subnet supported by the computing fabric 102, etc. Still further, any of the secured, encrypted connection techniques used within the NGPCAS 100 may be utilized in conjunction with endpoint authorization and authentication techniques to thereby provide even greater security for the NGPCAS 100.
Still further, in some implementations and as discussed above, in addition to or instead of using VPNs, the one or more transport networks 130 may utilize one or more point-to-point private connections (such as point-to-point and/or peer-to-peer Ethernet connections over private enterprise networks, and/or other types of secured, encrypted PTP, P2P, MTP, and/or multi-point connections) to securely deliver information between components of one or more NGPCAS systems 100. For example, point-to-point private connections can be established between two components 138, 140, between a physical site 115, 118 and the compute fabric 102, etc. For ease of discussion herein, though, the description refers to “VPN” techniques (not for limitation purposes) but to generally and categorically describe secured transport over the networks 130, although it is understood that any of the systems, methods, and techniques described herein may additionally or alternately utilize other types of secured, encrypted point-to-point, peer-to-peer, and/or multi-point connections for secured transport.
A user 155 of the NGPCAS 100 may authenticate to a VPN 158 in order to interact with the NGPCAS 100 (e.g., to interact with components 140, 135/138 of the NGPCAS 100) to, for example, view or obtain data and/or other information, change parameter values, configure or modify configurations of control routines and other aspects of the system 100, etc. As utilized herein, a “user” 155 may be a person or human user, or a user 155 may be an application or service executing on an external computing device that is not included in the system 100, e.g., an automated user. The users 155 depicted in
A human user 155 may be a person who is an agent of the enterprise that owns, manages, operates, or otherwise is associated with the NGPCAS 100. For example, a user 155 may be an enterprise configuration engineer, system operator, asset manager, supply chain manager, project or product manager, technician, installer, authorized third-party (e.g., a contractor), business manager, etc. Typically, a human user 155 interacts with the NGPCAS 100 via a computing device operated by the user (e.g., laptop, tablet, mobile computing device, in-vehicle computing device, etc.) on which an application, web browser, or similar executable executes to provide both a user interface operable by the human user 155 and a secure communicative connection with the NGPCAS 100 via a VPN 158. For instance, to utilize his or her computing device to interface with the NGPCAS 100, a human user 155 may utilize a particular MEEE 140 that has been configured and authorized to execute at the user-operated computing device (e.g., an app that has been downloaded from the compute fabric 102), or the human user 155 may utilize a web browser executing at the user-operated computing device. In another example, a user 155 may be an automated user such as an external application or service, e.g., an application or service executing on an external computing device or system. An automated user 155 may not provide any user interface for human users, yet nonetheless may establish a communicative connection with the NGPCAS 100 via the VPN 158 to obtain or provide data and/or other information. Examples of automated users 155 include third-party generated applications, external data sources (such as weather data sources, materials management systems, etc.), and the like. In some situations, an automated user 155 may be a particular containerized component or MEEE 140 that has been configured and authorized to execute on a remote computing device or system.
At any rate, irrespective of whether the user 155 is a human user or an automated user, the user 155 utilizes (via the VPN 158 to which the user 155 has authenticated) an API (Application Programming Interface) 160 to securely access data and/or other information at the NGPCAS 100. In an example implementation, different APIs 160 may be utilized by a user 155 to access different containerized components/MEEEs/granules 140 and physical components 135/138 of the compute fabric 400. In an additional or alternate implementation, a single API 160 may be utilized by the user 155 to access the NGPCAS 100, and the API 160 may form communicative connections, within the NGPCAS 100, with containerized components/MEEEs/granules 140 and physical components 135/138 as necessary to obtain or provide desired data and/or information to/from the user 155. The one or more APIs 160 may themselves be containerized components/MEEEs/granules 140 of the compute fabric 102, for example.
A particular user 155 of interest that is depicted separately in
In an embodiment, different subsets of containerized components/MEEEs/granules 140 may communicate with a particular physical component 135/138, a particular location 115, a particular set of multiple locations, an entirety of all of the physical locations of an enterprise, or respective physical components and/or respective physical locations of multiple, different enterprises via respective VPNs 162. For example, containerized components/MEEEs/granules 140 that are exclusively utilized by the architecture provider/manager 161 may communicatively connect with other containerized components/MEEEs/granules 140 and/or physical components 135/138 of an enterprise via a different, provider-specific VPN 162 than the VPN(s) utilized by the other containerized components/MEEEs/granules 140 of the enterprise to communicate with the physical components 135/138 of the enterprise location(s). In another example, the containerized components/MEEEs/granules 140 that are provided by the architecture provider/manager 161 may utilize a first enterprise-specific VPN 162 to communicate with containerized components/MEEEs/granules 140 and physical components 135/138 of a first enterprise, and may use a different, mutually exclusive enterprise-specific VPN to communicate with containerized components/MEEEs/granules 140 and physical components 135/138 of a second enterprise (not shown). As such, the architecture provider/manager 161 is able to securely and independently administer and manage the respective resources of different portions of a single enterprise and/or of different enterprises in a highly secure manner. Indeed, by using one or more of the VPN techniques described herein, either alone or in combination, the security of the NGPCAS 100 (and in some situations, of multiple NGPCASs 100 supported by the architecture provider/manager 161) may be customized as desired or needed.
Thus, in view of the above discussion, containerized components/MEEEs/granules 140 provided by the compute fabric 102 of the NGPCAS 100 include run-time logic functionalities (e.g., control and automation logic, data acquisition, etc.) that traditional process control and automation systems typically implement at Levels 1 and 2 of the Purdue model. The containerized components/MEEEs/granules 140 also include other logic functionalities that are related to the run-time logic and that traditional process control and automation systems typically implement at Levels 2 and 3 of the Purdue model, such as: system and component deployment; engineering; configuration; provisioning; commissioning; security of the system, devices, applications/services, and users; safety logic and systems; networking; monitoring; analytics; maintenance and diagnostics of the industrial process and equipment/assets; simulation; testing; fault and performance degradation detection and repair/recovery; operator interfaces; redundancy, backup, and other functionality related to availability of the system 100 and its components and equipment; historization of data; regulatory compliance; management of manufacturing or production workflow, execution, and operations; etc. Further, the containerized components/MEEEs/granules 140 may include logic functionalities that are typically provided in traditional process control systems at the higher Levels 4-5 of the Purdue model, such as enterprise resource planning, production scheduling, material use, shipping, inventory levels, and other enterprise-level functionalities. Still further, such containerized components/MEEEs/granules 140 can include logic functionalities that are introduced into the compute fabric 102 via third parties, such as applications and/or services that have been authored by third-parties, and that have been approved and authorized by the enterprise to be utilized within the NGPCAS 100.
Further, the set of containerized components/MEEEs/granules 140 provided by the compute fabric 102 may include containerized components/MEEEs/granules 140 of a networking layer (not shown in
Still further, the set of containerized components/MEEEs/granules 140 provided by the compute fabric 102 may include still lower-level logic functionalities (such as calculations, utilities, primitives, and the like) which may be utilized by containerized components/MEEEs/granules 140 at the application layer and at the networking layer of the compute fabric 102. Examples of such functionalities include computational functions (e.g., data aggregations and/or manipulations, such as averages, maximums, minimums, etc.); more complex calculations or algorithms (e.g., principal component analysis (PCA), partial least squares (PLS) predictions, and other types of statistical calculations and/or analytics); and process control-specific or automation-specific calculations (e.g., function block, shadow block, control module operations, etc.), to name a few.
Additionally,
Therefore, in view of the above, the compute fabric 102 of the NGPCAS 100 provides process control and automation functionalities and related functionalities that are required to be performed in traditional process control systems by different equipment and systems dispersed across Purdue Levels 1-5, and also provides additional, lower-level functionalities (e.g., from networking and platform resource management to calculations and primitives) which are available system-wide. Further, and advantageously, the NGPCAS 100 removes the necessity of DMZs at Level 3.5 and at other levels as well as eliminates the need for firewalls, data diodes, and other architectural security equipment and mechanisms utilized in traditional process control and automation systems, while simultaneously providing a more secured system 100.
In particular, as discussed above, any data or information transfer between two components (e.g., two different containerized components/MEEEs/granules 140, or a containerized component/MEEE/granule 140 and a physical component 125/135) of the NGPCAS 100 may be accomplished via a private session over a VPN, for example, a VPN utilized by the compute fabric 102 and a physical location at which physical components 102 are located, via a private VPN that is utilized by only the endpoint components, and/or one or more other types of VPNs. Further, for security purposes, any and all users 155/160 may be required to access the system 100 via one or more APIs 162/165 and a private VPN 158/162 established between the user 155/160 and the compute fabric 102. Still further, all users may be required to be multi-factor authenticated prior to gaining access to the system 100 via the APIs 162/165. In this manner, system data is exposed to only non-public addresses (e.g., addresses of components, compute fabric nodes, etc.) via VPN and authorized, credentialed entities. Applications may be exposed as websites or services, so that computing devices utilized by users access the system 100 via “thin” clients and do not have any system-related software (aside from, perhaps, the thin client such as a portal application) executing thereon. Further, all applications may be containerized, including the applications which execute locally at physical locations at which physical devices are disposed, and any system data may be encrypted at rest.
Additionally, within the NGPCAS 100, all communications transmitted and received over the network(s) 130 may be signed and encrypted, and plaintext protocols may be prohibited from use. Further, each NGPCAS 100 may have a single certifying authority (which can be associated with the enterprise, for example), and self-signed certificates and/or other forms of custom authentications may be prohibited.
As illustrated in
On the other hand, as illustrated with respect to the physical location 118A of
In both examples of
As mentioned above, in some implementations of the NGPCAS 100, the compute fabric 102 may facilitate the use of “digital twins” of one or more of the on-premises physical devices 105, 108.
Ethernet connections 208A-208D respectively couple the media converters 206A-206D to the compute fabric 102 and, in particular, to respective digital twins 210A-210D of the on-premises physical devices 202A-202D. The Ethernet connections 208A-208D may each connect directly to the compute fabric 102 (as depicted in
Each of the digital twins 210A-210D is, as the name implies, a “virtual twin” of the on-premises physical device, devices, or group of devices to which it is coupled (and which it represents). That is, each digital twin mimics the electronic functions, but not physical functions such as sensing and actuating, of one or more sensors, transmitters, actuators, etc. A digital twin may correspond to a single device (e.g., a pressure transmitter, a temperature transmitter, an actuator, etc.), may correspond to a group of devices (e.g., a valve having upstream and downstream pressure transmitters and a valve position sensor), or may correspond to a device class (e.g., a mass flow device).
Referring again to
In the contemplated systems implementing the compute fabric 102, the digital twins 210A-210D may be implemented as microservices (or other micro-encapsulated execution environments) executing, in embodiments, in joint or individual containers. In an embodiment, each digital twin 210A-210D self-identifies, within the system 100, using the unique network identifier or address of the respective physical device 202A-202D that the digital twin 210A-210D mirrors. Thus, other containerized components of the compute fabric 102 may communicate with an intended physical device 202A-202D by using the physical device's unique identifier, and may remain ignorant or unaware of whether the entity with which it is communicating is the actual physical device 202A-202D or its digital twin 210A-210D. As such, in a sense, each digital twin 210A-210D can serve as a proxy within the system 100 for the respective physical device 202A-202D that the digital twin 210A-210D mirrors. In further embodiments, and in a manner similar to that of other containerized components/MEEEs/granules 140 of the compute fabric 102, containerized digital twins may be nested according, for example, to area of the process plant, type of device, associated control module, etc. Of course, the AI, AO, and control algorithm modules may similarly be executed as containerized services or microservices.
The digital twins 210A-210D may, in embodiments, be created automatically from the descriptions of the respective physical devices 202A-202D they represent.
There are a variety of advantages that can be realized using a digital twin scheme as described here and depicted in
Another potential advantage of the use of digital twins is that it may contribute to the ability of the system described herein to be commissioned with less effort, expense, and time. By way of example and not limitation, a new plant (or portion thereof) identical to an existing plant (or portion thereof) may be instantiated almost entirely in the compute fabric, with the physical devices 202A-202D being coupled thereto by Ethernet. If the system, other than the physical devices 105, 108, is instantiated in the compute fabric 102, an entire process control plant (or at least elements in the compute fabric 102) can be instantiated within minutes, and automatic discovery (e.g., using device tags, including hardcoded device tags) can determine which devices to connect to the various containerized applications, containerized services, microservices, MEEEs, or granules instantiated in the compute fabric 102.
Still another advantage of employing digital twins is that the digital twins, connected directly to the control algorithms or not (i.e., whether the control algorithms interfaced with the digital twins or directly with the physical devices (e.g., 202A-202D) in the process plant) may be used to generate predictions as to the future state of the process and/or the physical devices. These future predictions could be generated using statistical models, mechanistic models, or a combination of statistical models and mechanistic models. These models, executing within the digital twin (e.g., as a nested containerized component/MEEE/granule), could receive the data from the physical devices and the control algorithm and provide assessments of the control system (e.g., the control system trending toward an abnormal state), the field devices (e.g., the sensor appears likely to fail or to have failed), etc. This information, in turn, could be used to provide to the operator, maintenance personnel, or others recommended mitigating or remedial actions.
Yet another advantage conferred by the use of digital twins is the introduction of a new source of soft-sensor data. Just as groups of physical devices (e.g., upstream and downstream sensors and a valve position sensor) may be represented by a single digital twin of the valve that embodies those three elements, groups of measurements (acquired by respective physical devices) may be collected by a digital twin and used as a “soft sensor” that gives an indirect measurement of another value. For example, a digital twin may receive values from four transmitters—two pressure transmitters together providing a differential pressure measurement, a pressure transmitter, and a temperature transmitter—and combine those into a soft sensor measurement for mass flow.
Further, digital twins as described above may also be implemented in traditional process plant environments. For example, an on-premises computing device may instantiate digital twins of the various physical devices in the process plant, even if the process plant implements traditional I/O and controllers. One such example arrangement is depicted in
The Next Generation Process Control and Automation System (NGPCAS) 100 described in the foregoing sections of this disclosure provides a number of enterprise-level benefits for any particular enterprise utilizing one or more NGPCASs 100 to implement industrial and/or automation processes. Examples of such benefits will be discussed herein with respect to
In some embodiments, compute fabric functionalities 320.1, 320.2 are executed in different portions of a shared compute fabric (e.g., sharing at least some compute fabric nodes between the enterprises 302, 304), with the security to the compute fabric functionalities 320.1, 320.2 being secured to the respective enterprises 302, 304 using security mechanisms of the present disclosure (e.g., even if components of compute fabric functionalities 320.1, 320.2 execute on a same compute fabric node, such components run as separate containerized components, with security and access being managed independently for each separate containerized component). At least a portion of compute fabric functionalities 320.1, 320.2 may be accessed and operated by credentialed users/computing devices associated with respective enterprises 302, 304 (e.g., human or automated users 155 and/or the architecture provider/manager 161 of
Similarly, the NGPCAS architecture of the present disclosure enables real-time monitoring (334) and operations functionalities (336) of an NGPCAS to be performed from multiple physical locations. That is, just as one operator or technician can remotely monitor the operation of multiple NGPCASs, various personnel across different physical locations may remotely monitor the operations of a single NGPCAS (or of a single physical device location). In embodiments, operators or technicians at operator workstations located at first, second, third, etc. physical device locations or even at other locations that are unassociated with any physical device location may monitor and perform operations associated with the run-time of portions of an industrial process (e.g., devices, groups of devices, control loops, etc.) executed at different ones of the first, second, third, etc. physical device locations. The workstation may be located at any physical location, e.g., at the physical device location at which the monitored/operated process executes, at a different physical device location, at a dedicated operations center, at a home office, in a vehicle, etc., and the workstation may be a stationary or a mobile device. The operator or technician may, for example, monitor the run-time operations of an industrial process by receiving, at the operator workstation via the compute fabric, various information associated with operation of the process (e.g., device configuration information, operator displays, process status information, equipment status information and operational parameters, process measurements, setpoints, alerts, and/or other information associated with components 135, 138 of
The NGPCAS architecture of the present disclosure may further enable control of an NGPCAS process or portion thereof (e.g., a device, group of devices, control loop, etc.) from any physical location (340). Configuration and execution of control loops may be implemented as containerized components in the location-agnostic NGPCAS compute fabric, rather than needing to be implemented on-premises at a physical device location of the controlled operation(s) as in traditional process control architectures. Portions of a control system may be distributed among containerized components executing at the same physical device location, a different physical device location, and/or other physical locations not associated with physical implementation of the process.
Indeed, any containerized component in the compute fabric (e.g., containerized services and applications) may execute at any physical location (338), e.g., at a physical device location to which the containerized components pertain, at a different physical device location, and/or at still other physical locations independent of any physical devices implementing at least a portion of the process. Similarly, containerized components in the compute fabric may be instantiated from any physical location (342). Containerized components may, for example, be instantiated via human and/or automatic operations at a first physical location, while the instantiated containerized components run or execute on compute nodes (of the compute fabric) that are located at a second physical location. Moreover, execution of containerized components may transfer on-demand or automatically between various physical locations (344, e.g., via the Orchestration service 422 of
Access of an NGPCAS to particular containerized applications/services can be provided on a per-application or per-service basis (“a la carte,” 348). In other words, an enterprise (e.g., one or more agents of the enterprise) can individually select particular monitoring applications, operational applications, control applications, dashboard applications, control modules, diagnostic applications, and/or other process functionalities to implement in the enterprise's NGPCAS to support the operation of a process. Instances of containerized components to implement the selected functionalities may be instantiated and executed on-demand or on an as-needed basis to support the scale of operation required by the enterprise NGPCAS.
In some embodiments, an architecture provider/manager (e.g., 161 in
Referring again to
Further to providing centralized management of application/service upgrades, the architecture provider/manager (e.g., 161 of
Still other enterprise-level benefits to the NGPCAS architecture of the present disclosure will be realized from the other portions of the present description.
With regard to the compute fabric 102 of the Next Generation Process Control and Automation System 100 of
Generally speaking, and as is described below, the compute fabric 400 utilizes a layered architecture in which business logic of the compute fabric 400 is abstracted from the physical computing platform of the compute fabric 400. For example, the compute fabric 400 may utilize one or more techniques and/or features described in U.S. patent application Ser. No. 17/487,609 filed on Sep. 28, 2021 and entitled “Software Defined Process Control System and Methods for Industrial Process Plants,” the disclosure of which is entirely incorporated herein by reference. For ease of discussion herein, the compute fabric 400 is described with simultaneous reference to the system 100 of
As shown in
Physical Layer of Compute Fabric
As further shown in
Each cluster Cx includes a plurality of nodes Ny that are communicatively interconnected to each other. Further, different clusters Cx may be physically disposed at the same or different physical locations (e.g., different locations 115, 118 of the NGPCAS 100 at which physical devices 105, 108 are disposed, and/or at one or more other locations at which no physical devices of the system 100 are disposed). A particular cluster Cx may be implemented solely at a single physical location or across multiple physical locations. Additionally, each data center cluster C1, C2, . . . , Cn is communicatively connected or networked with one or more of the other data center clusters C1, C2, . . . , Cn of the computing platform 405.
It is noted that although the physical layer 405 associated with the compute fabric 400 is described above as being implemented by using physical data center clusters C1-Cn, in some embodiments, at least a portion of the physical layer 405 may be implemented as a virtualized physical layer 405. For example, the data center clusters C1-Cn (or subset thereof) may be implemented as virtual machines, e.g., which execute on a computing resource platform, such as a cloud computing system.
Software Defined Networking Layer of Compute Fabric
The example architecture of the compute fabric 400 also includes a software defined (SD) networking layer 410 that interfaces the physical layer 405 of the compute fabric 400 with the software defined application layer 412 of the compute fabric 400. Accordingly, the software defined networking layer 410 is interchangeably referred to herein as the “operating system (OS) 410” of the compute fabric 400. Generally speaking, the OS 410 of the compute fabric 400 may assign, designate, or allocate various compute fabric nodes Ny to perform respective roles or functions to support the compute fabric 400, such as computing (e.g., via the nodes' respective processors and/or processing cores) or data storage (e.g., via the nodes' respective memories). Compute fabric nodes Ny that are assigned, designated, or allocated to perform computing activities of the compute fabric 400 are respectively referred to herein as “compute nodes” or “computing nodes.” Similarly, compute fabric nodes Ny that are assigned, designated, or allocated to perform storage activities of the compute fabric 400 are respectively referred to herein as “storage nodes.” An individual node Ny may be utilized as only a compute node, as only a storage node, or as both a compute node and a storage node, and the role(s) of each individual node Ny may dynamically change over time, for example, as directed by the OS 410. Advantageously, the computing platform 405 is scalable, so that individual nodes Ny and/or individual clusters Cx may be easily added, removed, swapped out, etc. as needed to support the compute fabric 400, and in particular, in accordance with the requirements of the other, higher layers of the compute fabric 400. For example, different nodes Ny of the compute fabric 400 may be assigned and re-assigned to different clusters Cx, and/or different nodes Ny and/or different clusters Cx may be physically disposed at different physical locations 115, 118 of the NGPCAS 100, as desired.
The operating system 410 of the compute fabric 400 executes on the computing platform 405 and, in an embodiment, may be built based on any suitable general purpose Hyper Converged Infrastructure (HCI) operating system (OS) such as Microsoft Azure Stack, VMWare HCI, Nutanix AOS, Kubernetes Orchestration, including Linux Containers (LXC/LXD), Docker Containers, Kata Containers, etc. As such, the OS 410 provides a set of computing, storage, and networking support services in a manner somewhat similar to general purpose HCI operating systems. However, in contrast to general purpose HCI OSs, and advantageously, in the compute fabric 400 of the Next Generation Process Control and Automation System 100, the OS support services are dynamically responsive to a logical or abstracted process control or automation system and other software components provided by the software defined application layer 412 of the compute fabric 400. That is, as performance, resource needs, and configurations of the various application layer services, subsystems, and other software components of the application layer 412 dynamically change (and/or are dynamically predicted, by services within the application layer 412, to change), the operating system 410 may automatically and responsively adjust and/or manage the usage of hardware and/or software resources of the physical layer 405 to support the needs and the requirements of the application layer 412 for computing, storage, and networking, as well as for other functionalities related to industrial process control and automation. To this end, the compute fabric operating system 410 may include a set of support services including, for example, a Software Defined (SD) computing (or compute) service 415, an SD storage service 418, an SD networking service 420, an SD orchestration service 422 (also interchangeably referred to herein as an “Orchestrator 422”), and optionally one or more other process control- and/or automation-specific SD OS support services and/or functions 425. For example, the process control- and/or automation-specific SD OS support services and/or functions 425 may include compute fabric individual resource and/or resource group management services that manage individual resources and/or groupings of resources provided by the software defined networking layer 410 and/or by the physical layer 405 of the compute fabric 400 such as, for example, virtual machines, containers, networks, network security groups, clusters, servers, and the like. As such, in an embodiment, the operating system 410 of the compute fabric 400 includes a general purpose HCI operating system platform (e.g., Microsoft Azure Stack, VMWare HCI, etc.) that has been particularly customized to include the SD computing service 415, the SD storage service 418, the SD networking service 420, the SD orchestration service 422, and the other process control and/or automation SD OS support services and/or functions 425, where the set of SD support services 415-425 is automatically responsive to and particularly supports application layer software components 412 of the compute fabric 400, that include, as previously discussed, process control- and/or automation-specific applications and services.
Interface Between Software Defined Networking Layer and Application Layer of Compute Fabric
In particular, as the compute fabric operating system 410 manages the allocation of the hardware and software resources of the nodes Ny of the computing platform 405 via the SD OS support services 415-425, the SD OS support services 415-425 may also serve as interface services between the OS 410 and the higher level services, subsystems, and other software components of the application layer 412 of the compute fabric 400, and/or may provide a framework for these higher level services, subsystems, and other software components of the application layer 412. As such, the software components of the compute fabric application layer 412 may interface with the OS 410 (and in some cases, particularly with one or more of the SD-specific support services 415, 418, 420, 422, 425 provided by the OS 410) via a set of Application Programming Interfaces (APIs) 428, either via an HCI Adaptor 430 (also referred to herein as an “HCI Adaptor layer” 430) and another set of APIs 432, or directly (not shown in
Thus, unlike generalized, layered IT (Information Technology) system architectures in which business logic applications are abstracted from the hardware and software computing platform and for which the management of computing platform resources is largely governed and designed by human IT administrators, the architecture of the compute fabric 400 not only abstracts the higher level, business logic services, subsystems, and other software components of the application layer 412 from the hardware and software computing platform 405, but also enables the higher level software defined services, subsystems, and other software components 412 to dynamically, automatically, and responsively direct and cause changes to the usage of the hardware and software resources of the nodes Ny and clusters Cx of the physical layer 405 and of the software defined networking layer 410 (and optionally, of groups of hardware and/or software resources thereof), e.g., via the APIs 428 and the SD OS support services 415, 418, 420, 422, 425, and without requiring any human intervention or direction. Particularly, and advantageously, the management of the resources of the physical layer 405 and of the software defined networking layer 410 is dynamically responsive to changes in the configurations and needs of these higher level SD services, subsystems, and other software components of the application layer 412, and in particular, with respect to the particular requirements, metes, and bounds of industrial process control and automation systems, such as timing, synchronization, and/or other control- or automation-specific constraints.
Containers and Other Types of Micro-Encapsulated Execution Environments
As shown in
The application layer software components 435-448 of the application layer 412 may execute in containers and/or in other suitable types of micro-encapsulated execution environments (MEEEs) or granules, e.g., as Instantiated Software Components (ISCs). For example, an ISC may be a container configured with an instance of a particular application layer software component 435-448 to form a configured container, container image, or other type of micro-encapsulated execution environment or granule for the particular application layer software component 435-448, and the container image of the particular application layer software component 435-448 may be instantiated for execution on a particular compute fabric node Ny as a specific instantiated MEEE or ISC. Said another way, a configured container may be an instance of an application layer software component 435-448 that is configured into a respective container or other type of micro-encapsulated execution environment or granule.
Generally speaking, containerized or micro-encapsulated software components or MEEEs/granules (e.g., at the application layer 412, the HCI adaptor layer 430, and the software defined networking layer 410) are included in the set of containerized/micro-encapsulated components 140 of the NGPCAS 100 and, as such, are isolated from other containerized/micro-encapsulated services and applications (e.g., other containerized/micro-encapsulated components 140) that are executing on the same node Ny. As such, the terms “configured container,” “container image,” and “containerized component” are utilized interchangeably herein and, for ease of discussion but not for limitation purposes, are generally and categorically utilized herein to refer to any one or more suitable types of instantiated, micro-encapsulated execution environments or granules, such as software containers, virtual machines, software agents, scripts, functions, calls, actors (e.g., lightweight processes such as Erlang, Scala, Akka, etc.), unikernels (e.g., machine images which run on bare metal and contain all components necessary to execute an application, including the operating system component), other types of bare metal software (e.g., software running directly on hardware without any intermediate managing software, such as a hypervisor or other type of container/encapsulation manager), and/or other types of micro-encapsulated execution environments or granules.
Various MEEEs or granules can be configured to execute (e.g., when instantiated) various broad to granular operations within the NGPCAS 100. To illustrate using an example, a control routine may include multiple control modules which operate cooperatively to execute the control routine, a control module may respectively include multiple function and other types of blocks which operate cooperatively to achieve the control module, and a function block may respectively include multiple granular operations which operate cooperatively to execute the function block. Accordingly, in one implementation of this example, a single MEEE can be configured to execute (e.g., when instantiated) an entirety of the control routine. In another implementation of this example, each MEEE of a group of MEEEs can be respectively configured to execute (e.g., when instantiated) a different control module or portion of the control routine, and the group of instantiated MEEEs may operate cooperatively to thereby execute, as a group, the control routine in its entirety. Indeed, in some implementations, a second group of MEEEs can be respectively configured to execute (e.g., when instantiated) different granular operations or portions of a control module (such as an input function block, an error detection function block, a control function block, a logic function block, a script, an output function block, etc.), and this second group of MEEEs, when instantiated, may operate cooperatively to thereby execute, as a group, the control module in its entirety. In still other implementations, a single MEEE may be configured to execute (e.g., when instantiated) as a process controller, a process control subsystem, a unit, an area, or even the process control system 445 as a whole.
In another example, a single, individual MEEE can be configured to execute (e.g., when instantiated) an entirety of a complex data analytics routine for the entire NGPCAS 100. Alternatively, each MEEE of a group of MEEEs can be respectively configured to execute (e.g., when instantiated) a different simple data analytics routine (or some other respective portion of the complex data analytics routine), and the execution of the instantiated group of MEEEs in cooperation may thereby cause the execution of the entirety of the complex data analytics routine. In some implementations, another group of MEEEs, when instantiated, can execute respective granular actions or operations of a simple data analytics routine (or other types of respective granular actions of a simple data analytics routine) in cooperation to thereby cause the execution of the entirety of the simple analytics routine (or the entirety of the portion of the complex data analytics routine, as the case may be). For instance, granular analytics actions or operations may include computational functions (e.g., data aggregations and/or manipulations, such as averages, maximums, minimums, etc.), simplex data analytics routines may include more sophisticated statistical calculations or algorithms (e.g., principal component analysis (PCA), partial least squares (PLS) predictions, and other types of statistical calculations and/or analytics), and complex data analytics routines may include combinations of statistical calculations or algorithms, and in some cases in combination with other type of non-statistical calculations or algorithms.
Generally speaking, MEEEs, granules, or configured containers may be provided by the enterprise (e.g., via the architecture provider/manager 160, via an application/service store, out-of-the-box, etc.), users 155 who are agents of or associated with the enterprise 155, third-parties, and/or other sources. Various instantiated MEEEs may be assigned to execute on various compute nodes Ny of the system 400, which may be disposed in different physical and/or geographical locations. Further, the instantiated MEEEs may be dynamically migrated from executing at one node to executing at another node, e.g., based on detected and/or predicted resource usage, jurisdictional requirements and/or regulations, and/or other criteria. Still further, instantiated MEEEs may be allocated, pinned, dynamically assigned, and/or dynamically migrated, e.g., via the SD compute service 415, to execute on respective nodes Ny and/or data center clusters Cx. The SD compute service 415 may dynamically change, update, maintain, and/or otherwise manage the container images and their respective assignments to compute nodes Ny as or when needed, e.g., to load-balance across compute nodes Ny, for scheduled maintenance of compute nodes Ny and/or physical components thereof, in response to detected and/or predicted resource usage and/or performance issues, to support expansion or contraction of the logical process control or automation system 445, to support expansion or contraction of the computing platform 405, based on jurisdictional regulations and/or requirements, etc. As such, the NGPCAS 100 may be viewed as a dynamic, highly-distributed set of MEEEs or as a dynamic mesh of MEEEs, where the MEEEs may be located or disposed across multiple physical and/or geographical locations, and where one or more of the MEEEs may be dynamically reassigned and migrated, during run-time operations of the NGPCAS 100, to another node and/or another physical and/or geographical location while maintaining the execution of the run-time operations of the NGPCAS 100. Moreover, components of the dynamic meh of MEEEs forming a particular application, control routine, analysis routine, etc. may be dynamically moved or migrated or reassigned to other hardware in the compute fabric individually, in sets or groups, or altogether if desired without effecting or interrupting operation of the application, routine, etc. As such, individual components of the dynamic mesh of MEEEs for a particular application or usage may be managed in the compute fabric separately from other components of the same application or usage.
Within the compute fabric 400, some configured containers, granules, or instantiated MEEEs may be allocated or assigned to respective compute nodes Ny and dynamically re-assigned to different compute nodes Ny by the SD compute service 415 based on dynamically changing configurations, performance, and needs of the logical process control or automation system 445. In some situations, configured containers may be assigned (and re-assigned) to be executed by particular processors or particular processor cores of SD compute nodes Ny. Some configured containers, though, may be pinned to respective SD compute nodes Ny (e.g., by the SD compute service 415, by a configuration, by a user, etc.) and are not dynamically re-assigned by the SD compute service 415 due to dynamically occurring conditions. That is, a pinned configured container may execute on the compute node Ny to which the configured container is pinned until the configured container is unpinned from the compute node Ny, e.g., irrespective of dynamic conditions of the logical process control or automation system 445 (other than perhaps the failure of the compute node Ny to which the configured container is pinned). Said another way, the software defined networking layer 410 may limit the utilization, by the pinned configured container, to only the hardware and/or software resources to which it is pinned, and when the configured container is unpinned, the SD networking layer 410 removes the limitation. Configured containers may additionally or alternatively be pinned to other physical or logical components of the compute fabric 400, if desired. For example, a configured container may be pinned to another configured container, to a specific data cluster, to a particular processing resource (e.g., a particular physical processor or a particular physical processor core of a compute node Ny), to a physical rack or portion of a physical rack serviced by a particular power supply (where the physical rack physically houses the hardware of one or more compute fabric nodes), etc.
Further, configured containers, instantiated MEEEs, or granules may be nested within and/or pinned to other configured containers, which is particularly useful in configuring and organizing the logical process control or automation system 445. For example, when a particular process control subsystem 438 provides a particular set of control services 435 and/or other services 440, a configured container of each provided service 435, 440 of the particular set may be nested in the configured container of the particular process control system 438. In another example, multiple control routine and/or control module configured containers may be nested within a specific controller service 435, and the specific controller service 435 may be nested within the particular process control subsystem 438. In yet another example, a controller or control service 435 may be configured with one or more process control module services 435, parameters, and values of the industrial process plant 10, such as tags of inputs and outputs, reference values, and the like, thereby forming a configured or programmed controller service. The controller or control service 435 may be functionally equivalent to a traditional, dedicated, hardware controller device as understood in the Purdue model, or the controller service 435 may be functionally equivalent to a control routine or control module which is configured into an executed by the traditional, dedicated hardware controller device. A container may be configured with an instance of the configured controller service, thereby forming a container image or instance of the configured controller service that is executable to perform the specific, configured set of process control logic, e.g., by using the configured control module containers, tags, reference values, etc., when so configured. Multiple instances or container images of a configured controller service (or of other configured applications and services) may be instantiated and executed by the compute fabric 400.
In still another example, containers, granules, or instantiated MEEEs within the SD application layer 412 may be utilized to represent and/or logically organize physical and/or logical areas, regions, and components of the NGPCAS 100. For examples, units, areas, and the like may be represented by respective configured containers, and configured containers corresponding to physical and/or logical components of each unit, area, etc. may be nested within and/or pinned to their respective configured, organizational container(s). As such, within the compute fabric 400, a configured control routine container may be nested within or pinned to the configured controller container, and the configured controller container may be nested within or pinned to another configured container, e.g., a container that has been configured for a depropanizer.
For clarity and ease of discussion herein, the term “container” is utilized herein to generally refer to an instantiated software component (ISC) that is a configured container, container image, containerized component, or other type of micro-encapsulated execution environment (MEEE) or granule, e.g., a container or other type of micro-encapsulated execution environment that has been configured to include an instance of a respective controller service, subsystem, or other service or application provided by the application layer 412 of compute fabric 400.
At any rate, and in a manner similar to that discussed for the computing resources of the computing platform 405, containerized/microencapsulated components 140 of the system 100 may be dynamically allocated and/or assigned, pinned, and/or nested, e.g., via SD storage service 418, to various compute fabric storage nodes Ny to thereby support various storage needs of the logical process control or automation system 445. For example, the SD storage service 418 may administer and manage the logical storage resources utilized by configured containers of the logical process control or automation system 445 across various physical hardware memory resources of one or more nodes Ny. For instance, the configured container and the memory needed for its operations (e.g., Random Access Memory or similar) may be stored on a particular SD storage node Ny or a particular memory device or space of an SD storage node Ny. Additionally, if desired, some containerized components/MEEEs/granules 140 may be pinned to respective SD storage nodes Ny and/or to specific memory devices or memory areas of the SD storage nodes Ny. The SD storage service 418 may change, update, or otherwise manage the physical hardware memory or memories of the computing platform 405 to support logical storage resources of the compute fabric 400 when and as needed, e.g., due to disk or other types of errors, for scheduled maintenance, due to the addition/expansion of available physical memory in the computing platform 405, etc.
Still similarly, the SD networking service 420 may administer and manage the logical or virtual networking utilized by the containerized components/MEEEs/granules 140 of the logical process control or automation system 445 and/or by other containerized components/MEEEs/granules 140, which may be implemented by the SD networking service 420 across the compute fabric nodes Ny. For example, the SD networking service 420 may administer and manage networking and hardware resources of the computing platform 405 to support the logical networking functionalities included in the logical process control system 445, such as virtual interfaces, virtual switches, virtual private networks, virtual firewall rules, and the like, as well as to support required networking between various configured containers or container images executing on the compute fabric 400. Further, as the logical process control system 445 services the physical components 135, 138 of the NGPCAS 100, the timing and synchronization of the containerized components/MEEEs/granules 140 of the compute fabric 400, the physical components 135, 138 of the field environment 120, and the networking there between is critically important, as missed and/or lost messages or communications may result in the industrial or physical process becoming uncontrolled, which may in turn lead to catastrophic consequences such as overflows, gas leaks, explosions, loss of equipment, and, in some situations, loss of human life. Fortunately, the SD networking service 420 is responsive to the critical process I/O timing and synchronization of the compute fabric 400 so that communications (and in particular, communications to/from control services 435), may be reliably delivered in a timely and deterministic manner. For example, the SD networking service 420 may support the time synchronization of data center clusters Cx to within 1 millisecond to ensure required synchronization between process control services 435, process control subsystems 438, the packet router/switch service 442, and other software defined services 440, 448 of the software defined application layer 412.
In addition to the SD compute service 415, the SD storage service 418, and the SD networking service 420, the compute fabric operating system 410 may provide other OS support services 425 that are accessible via the set of APIs 428, 432 and which may be utilized or accessed by the application layer 412 to support the logical process control system 445 and other containerized components/MEEEs/granules 140 of the application layer 412 of the compute fabric 400. For example, the other OS services 425 may include a service life cycle management service, a discovery service, a security service, an encryptor service, a certificate authority subsystem service, a key management service, an authentication service, a time synchronization service, a resource and/or resource group management service, a service location service, and/or a console support service (all not shown in
Indeed, in an embodiment, one or more of the software defined components 415-425 and 452-460 of the software defined networking layer 410 are implemented as respective configured containers or container images of the compute fabric 400. That is, one or more services and other functionalities provided at the software defined networking layer 410 of the compute fabric 400 (and in some implementations, all of the services and functionalities provided at the software defined networking layer 410) may be implemented as respective containerized components/MEEEs/granules 140 of the NGPCAS 100. As such, in manners similar to those discussed herein for containerized components/MEEEs/granules 140 of the application layer 412, containerized components/MEEEs/granules 140 of the software defined networking layer 410 may be uniquely identified within the NGPCAS 100 by a respective address, may communicatively connect to other containerized components/MEEEs/granules 140 of the NGPCAS 100 and optionally to the physical components 135, 138 of the NGPCAS 100, can be spun up and removed as or when needed, etc.
Application Layer of Compute Fabric
Now turning more specifically to the application layer 412 of the compute fabric 400, and as shown in
Each different control service 435 may be configured with desired parameters, values, etc. and optionally other control services 435; each instance of a configured control service 435 may execute in a respective container; and each configured container may be assigned (or pinned) to execute on a respective compute node Ny and/or cluster Cx. As such, each configured control service 435 may be a logical or software defined control entity which functionally may be configured and may perform in a manner similar to that of a traditional, hardware-implemented process controller device, control module, process control function block, etc. However, unlike traditional, hardware-implemented process controller devices, traditional control modules, and traditional control function blocks, and advantageously, the compute fabric 400 may easily replicate multiple instances of a same configured control service 435 for various purposes, such as performance, fault tolerance, recovery, and the like. For example, a controller service (that executes in its own container) may be configured to execute a control module service (that executes in its own container), and the control module service may be configured to execute a set of control function block services (each of which executes in its own container, and each of which may be configured with respective parameters, values, etc.). As such, the set of configured containers corresponding to the set of configured control function block services may (though need not necessarily) be nested in the configured control module service container, and the configured control module service container may be nested in the configured controller service container. The set of configured containers corresponding to the set of configured function block services may be assigned to execute on different cores of a particular processor of the compute platform 405, e.g., for performance load-balancing purposes. When loads change, one or more of the configured function block service containers may be moved to execute on different processor cores, different processors, or even different compute fabric nodes in attempt to re-balance loads; however, the moved function block service containers would still be nested under the configured control module service container, and would execute accordingly.
In addition to control services 435, other types of application layer services 440 related to industrial process control may be provided by the application layer 412, such as, but not limited to, operator displays and interfaces, diagnostics, analytics, safety routines, reporting, historization of data, configuring of services, configuring of containers, communicating information with external or other systems, enterprise-level applications, process control and/or automation resource and/or resource group management, etc. For example, a process control and/or automation resource group management service may allow a user 155 to group and/or isolate various resources based on the NGPCAS 100 and/or other process control or automation considerations. For instance, resource groups may be formed based on physical characteristics, such as sites or physical locations, groups of sites/physical locations, subsets of sites or physical locations, geographical regions, etc.; based on logical characteristics, such as categories, containers and/or container types, control strategies, capabilities, timing, performance, user characteristics, etc.; based on functionalities, such as storage items, networking items, etc.; and/or based on other types of groupings and/or combinations thereof corresponding to the NGPCAS 100. Generally speaking, any process control or automation system-related functionality or business logic that executes during run-time of the NGPCAS 100 to control the industrial process, supports the NGPCAS 100, and/or relates to the NGPCAS 100 may be logically implemented in the compute fabric 400 as a respective application layer service 435, 440 executing in a respective container. For example, any one or more of the enterprise-level compute fabric functionalities 320 may be implemented in respective containers or as containerized services. Further, any of the containerized services 435, 440 may communicatively connect, e.g., via the SD networking layer 410, with respective physical components 135/138 disposed in physical locations of the NGPCAS 100 when required to do so by the business logic of the service 435, 440 and/or by the recipient physical components 135/138. Still further, any of the containerized services 435, 440 may communicatively connect any other containerized service 435, 440 to transfer data and/or information there between when required to do so by their respective business logic.
In a similar manner, each different subsystem 438 of the application layer 412 of the compute fabric 400 may be provided by or execute in a respective container. The set of subsystems 438 provide the virtual or logical process control-related subsystems of the logical process control system 445. In some cases (not shown in
Examples of subsystems 438 which may be provided by the application layer 412 of the compute fabric 400 include, but are not limited to:
Further, the application layer 412 of the compute fabric 400 may include additional or alternate subsystems 438 which may be utilized by the system 100. For example, the subsystem 438 may include a control strategy subsystem directed to higher-level and/or overall control strategies, e.g., to achieve product and process performance and quality targets; an analytics subsystem; an optimization subsystem, a mass and energy balancing subsystem, a security subsystem, which may include one or more specialized algorithms for detecting security intrusions, etc.
Software Defined Router/Switch Service
The software defined packet router/switch service 442 generally operates as an application or service that is responsible for transferring packetized I/O data (and, in some scenarios, other types of packetized data or information) between endpoints of the NGPCAS 100, e.g., from a physical component 135/138 to a containerized component/MEEE/granule 140 at the application layer 412 of the compute fabric 400 and vice versa; from a physical component 135/138 to a containerized component/MEEE/granule 140 at the networking layer 410 of the compute fabric 400 and vice versa; from a containerized component/MEEE/granule 140 at the application layer 412 to another containerized component/MEEE/granule 140 at the application layer 412; from a containerized component/MEEE/granule 140 at the networking layer 410 to another containerized component/MEEE/granule 140 at the networking layer 410; from a containerized component/MEEE/granule 140 at the application layer 412 to a containerized component/MEEE/granule 140 at the networking layer 410 and vice versa, and the like. For example, the packet router/switch service 442 may communicatively couple the respective endpoints and transfer data there between using any suitable data delivery or data transfer paradigm, including request/response, publish/subscribe, etc. In an embodiment, when at least some of the software defined application layer software components 435-448 are deployed as microservices/MEEEs/granules that are communicatively connected by a microservice/MEEE/granule bus (not shown), the packet router/switch service 442 (in some cases, in conjunction with the OS 410 and its support services 415-425) may support and/or manage the microservices/MEEEs/granules and microservice/MEEE/granule bus so that the microservices/MEEEs/granules may transfer data and/or information there between (e.g., in a packetized format). In an additional or alternative embodiment, the software defined packet router or switch service 442 may use any one or more packet-based networks or links (such as the networks 402 and/or software defined links provided by the compute fabric 400) to transfer packetized data and information between one or more containerized components/MEEEs/granules 140 and/or physical components 135/138 of the NGPCAS 100.
It is noted that
Alternately, the packet router/switch service 442 may be an application layer software defined service 440 which is in its own, stand-alone subsystem within the application layer 412, or which is not associated with any subsystem. At any rate, and generally speaking, the software defined packet router/packet switch service 442 may be typically implemented as a containerized component/MEEE/granule 140 of the compute fabric 400.
As such, the containerized packet router/switch service 442 may be accessed by other containerized components/MEEEs/granules 140 of the compute fabric 400 (both at the application layer 412 and the networking layer 410) for the purposes of data transfer or data delivery. In some situations, the packet router/switch service 442 may utilize the APIs 428 to thereby cause the transfer of packetized I/O data and/or other types of packetized data, e.g., via the OS support services 415-425. In some situations, the packet router/switch service 442 may cause data to be transferred via the microservice/MEEE/granule bus. In effect, the packet router/switch service 442 serves as a logical or gateway (e.g., an API gateway) that causes packetized process I/O and/or other types of packetized data to be routed between configured containers of the compute fabric 400, and that causes packetized process I/O, packetized control signals or instructions, and other types of packetized information to be routed between configured containers of the compute fabric 400 and physical components 135, 138 deployed in the field environment 120 of the NGPCAS 100.
As will be understood, one significant advantage of the system described herein is that it reduces the data movement and data storage needed to support applications or other usages executing in in real-time in the compute fabric. In particular, because of the secured, real-time communication structure provided in the NGPCAS described herein, including the use of secured, encrypted data links (e.g., VPNs) between various software and hardware elements both in the compute fabric and in the physical locations (e.g., the plants), elements such as MEEEs in the compute fabric can access data from anywhere in the system (i.e., from any other element where that data was created or resides) in real-time. This feature means that applications and other application usages executed in or by higher level system elements or platforms (e.g., a control application, a maintenance application, a data logging or tracking application, a fleet management application, etc.), or any individual MEEE or granule thereof can access data wherever this data resides (e.g., in another MEEE or granule anywhere in the system, in a compute fabric database, in a physical location database or server, in a field device at a physical location, etc.) in real-time using, for example publish/subscribe communications, dedicated data calls, etc. In fact, individual MEEEs or other compute elements (granules) that make up or implement a particular application or usage can include a pointer or reference that points to the data it needs for operation wherever that data resides in the system and the granule can access that data in real-time when it is needed or used by the MEEE or granule. Thus, granules can, for example, access and use data wherever that data is created and/or initially stored. This feature means that data does not need to be moved from, for example, a server in a physical location to a cloud-based server in the compute fabric prior to being usable or accessible in real-time by an application or element (e.g., granule) in the compute fabric. This feature thereby speeds up computing operations and reduces data flow that was, in the past, performed simply for the purpose of moving data from one location to another to make that data available for use in real-time to applications which use that data. The system described herein thus enables data to be used wherever it resides or is generated by direct data calls or publish/subscribe communications, whether that data resides in or is generated by a device within the compute fabric or by a device at a physical location or plant.
Logical/Virtual Components
Further, at the application layer 412 of the compute fabric 400, at least some physical process control devices or components (e.g., controllers, safety logic solvers or devices, data storage devices, etc.) of traditional process control systems may be logically implemented in the logical process control system 445 as a respective service 435, 440 or subsystem 438 executing in a respective container. Such logical or virtual instances of process control devices or components may be configured in a manner similar to their physical counterparts, if desired, by configuring the logical devices with control routines, other application layer software components 412, parameters, reference values, lists, and/or other data. For example, a controller service may be configured with several control modules, a display view service may be configured with user access controls and graphical elements, etc. Configured logical or virtual process control devices or components (e.g., container images of process control devices or components) may be identified within the logical process control system 445 via a respective device tag or identification, for example, and respective signals that are received and generated by configured logical or virtual instances of process control devices may be identified within the logical process control system 445 via respective device signal tags or identifiers. A logical or virtual instance of a process control device may be uniquely identified within the system 100 and operate as an individual entity in lieu of any corresponding physical device of the system 100, or a logical or virtual instance of a process control device may be a proxy or digital twin of a physical device included in the system 100, such as previously described.
At the software defined application layer 412, the compute fabric 400 also includes software defined storage entities or components 413, which may provide abstracted data storage (and access thereto) for the services and subsystems 435-448 of the SD application layer 412. For example, historian databases, configuration databases, and other types of process control system databases and data storage entities as well as temporary storage utilized by various process control application services 435-448 during execution may be provided by the software defined storage entities 413. The storage databases, areas, devices, etc. may virtualized or logical storage entities or components, which may be assigned or allocated (and may be re-assigned and re-allocated) to various storage resources of the nodes Ny of the computing platform 405 by the compute fabric operating system 410. For example, a single software defined logical database may be implemented over the hardware memory resources of multiple nodes Ny. Additionally, the SD Storage service 418 of the compute fabric operating system 410 may assign/re-assign and re-assign/re-allocate software defined storage entities 413 at the application layer 412 to different storage resources provided by the nodes Ny based on performance, resource, and configuration needs of the storage entities or components 413 and optionally of other components of the SD application layer 412.
Orchestration
Returning now to the software defined networking layer 410 of the compute fabric 400,
To this end, the performance-related services 452-460 of the OS 410 may monitor performance parameters, resource usage, and/or criteria during run-time, detect any associated conditions which occur and/or which are predicted to occur, and provide and/or implement any changes in assignments of application layer software components (e.g., containerized components) 412 to hardware and/or software resources of the computing platform 405. Accordingly, during run-time of the system 100, as various expected and/or unexpected hardware and/or software conditions arise and are detected, the Orchestration service 422 responsively adjusts the allocation of hardware and/or software resources of various compute fabric nodes Ny to instantiated container images to maintain (or attempt to maintain) a target or best-effort level of performance and fidelity of operations. Detected conditions which may cause the Orchestration service 422 to modify allocations and/or assignments between containerized components 412 and physical resources of nodes Ny may include, for example, hardware faults or failures, software faults or failures, overloading of a specific compute fabric node, increased or decreased bandwidth of various networking components, addition or removal of compute fabric nodes and/or clusters of compute fabric nodes, hardware and/or software upgrades, pinning and/or unpinning of containerized services and/or applications, diagnostics, maintenance, and other routines which may cause hardware and/or software resources to be temporarily unavailable for run-time use, etc. Possible responsive and/or mitigating administrative actions which may be taken by the Orchestration service may include, for example, re-assigning containerized services and/or applications to execute using different software and/or hardware resources (in some cases, on different nodes Ny), activating and/or deactivating software and/or hardware resources, changing priorities of various containerized components' access to various software and/or hardware resources, etc.
Accordingly, and generally speaking, the services, subsystems, and other software components of the software defined application layer 412 (e.g., 435, 438, 440) may determine, define, or specify the processing, containerization, networking, and storage needs of the logical process control system 445, both at an individual container level and at aggregate levels (e.g., at a subsystem level, unit level, area level, and/or the process control system 445 as a whole). By way of the APIs 428 (and, in some configurations, also by way of the HCI adaptor layer 430 and APIs 432), the OS 410, its support services 415, 418, 420, 422, 425, and its Orchestration service 422 administer and manage the hardware and software resources of the compute fabric nodes Ny to support those needs. For example, in some embodiments, the SD Orchestrator 422 may cause different instances of a particular control routine 435 or of a particular other service 440 to execute on different nodes Ny, e.g., for fault tolerance, quality of service, and/or other performance criteria of the compute fabric 400. Advantageously, as the needs of the logical process control system 445 dynamically change over time, the OS support services 415, 418, 420, 422, 425 and/or the Orchestration service 422 may modify, change, and adjust the usage of the hardware and software resources of the nodes Ny, e.g., in a responsive and/or in a predictive manner.
For example, when the logical process control system 445 creates additional instances of control services 435 executing in additional containers, the OS support services 415-425 may responsively (via the APIS 428 and optionally the HCI adaptor 430 and the APIs 432) assign the newly created containerized components to execute on corresponding compute fabric nodes Ny, may re-balance existing containerized components among nodes Ny, may assign specific hardware memory resources to support the logical memory resource needs of the additional containerized components, may adjust routing tables utilized by the nodes Ny to support the logical routing needs of the newly created containerized components, etc. In another example, when a particular cluster C2 needs to be taken out of service (e.g., expectedly for maintenance purposes or unexpectedly due to a lightning strike), the OS support services 415-425 may pre-emptively re-assign containerized components that are presently assigned to execute on cluster C2 to other clusters in accordance with the present needs of the logical process control system 445 and the availability of hardware and/or software resources of the other clusters, and the support services 415-425 may adjust routing tables utilized by the clusters Cx accordingly so that continuity of execution of said containerized components is maintained even when the cluster C2 is taken out of service.
As such, the software defined networking layer 410 automatically, dynamically, and responsively determines, initiates, and performs changes to the allocation of hardware and software resources of the nodes Ny of the computing platform 405 to different application layer software components 412 based on detected conditions, such as improvement in performance of individual logical and/or physical components or groups thereof, degradation of performance of individual logical and/or physical components or groups thereof, fault occurrences, failures of logical and/or physical components, configuration changes (e.g., due to user commands or due to automatic re-configuration by services of the compute fabric 400), etc. Consequently, the compute fabric 400 may automatically redistribute hardware and software resources of the nodes Ny responsive to changing conditions and components of the compute fabric 400 to thereby support the process control system 245 and other services executing at the application layer 412 of the compute fabric 400.
Simulation
In some implementations, the compute fabric 400 may implement simulations of or changes to various application services 435, 440, 448, to the entire software application layer 412, to various support services 415-425, 452-460, and/or to the entire software defined networking layer 410. That is, a simulation of the target components/layers may execute in concert with the active software defined components/layers on top of the computing platform 405, and thereby receive run-time data from the field environment of the industrial process plant and operate accordingly, e.g., with the same logic, states, timing, etc. as the active target components/layers, or with the simulated test logic, states, timing, etc. However, the I/O and other types of data generated by the simulation are prevented from being delivered to the field environment, and simulations may be paused, sped up, slowed down, fed test inputs, and otherwise managed to observe behavior and make modifications to the simulated components/layers. Accordingly, upon approval of a simulated portion of the compute fabric 400, the simulated portion may simply be activated for use during run-time operations of the industrial process plant, without needing to pause or take part of the compute fabric 400 out of service to do so.
Next, various security features of the Next Generation Process Control and Automation System (NGPCAS) 100 are described. As previously discussed, present-day process control and automation systems that are designed around the Purdue model suffer from many drawbacks, including increased complexity (and thus, more opportunity for component and process failure), decreased performance, and greater risk of cyber-intrusion, to name a few. For example, in present-day process control and automation systems, typically at least three different domains may exist between Levels 2, 3, and 4, and the security policies for each domain may be different and require different security management techniques. As such, cross-level connectivity is challenging to implement, as well as can introduce significant delays for the delivery of data across multiple Purdue levels, e.g., from Level 2 to Level 4 and above. Further, industries are looking towards being able to deliver instructions, commands, and/or other information from higher levels of the Purdue model to lower levels. For example, a technician or other process plant personnel may be working remotely via his or her portable computing device, and may want to monitor run-time process plant operations and adjust configurations, parameter values, settings, etc. within the process control system in response to monitored conditions. In such situations, when instructions and information move from higher levels of the Purdue model to lower levels (e.g., via “holes” that are added to the established security mechanisms for these purposes), outbound firewalls and other currently implemented security mechanisms that are designed and utilized to prevent the inflow of external information into the plant must necessarily be compromised, thereby introducing significant additional risk of external parties accessing the information and data in the protected lower levels of the plant and other types of cyber-intrusion. Moreover, the multiple security mechanisms implemented between Purdue layers, either as initially designed or with any added “holes,” create a highly complex network, which is difficult to engineer, maintain, and utilize to efficiently deliver information between Purdue levels.
Further, in present-day systems that perform some process control and/or automation functionality in the cloud, other undesirable issues are introduced. For example, when the process plant of such systems loses Internet connectivity, the cloud cannot be accessed and any control or automation functionality provided by the cloud is unavailable. Additionally, still further latency, bottlenecks, and complexity are added with cloud-based implementations beyond those introduced by Purdue-model implementations. Moreover, sufficiently secure mechanisms for supporting native communication between process control devices (e.g., field devices) and the cloud do not exist.
The security features of the NGPCAS 100 address at least these known security issues of Purdue model implementations as well as provide additional security, improved performance, easier engineering and maintenance of the NGPCAS 100, as well as other benefits and advantages. Examples of such security features are described below. Any of the described security features may be used as a stand-alone security feature or in combination with any other one or more security features. In some embodiments, various security features of the NGPCAS 100 may be implemented as services 425 provided by the software defined networking layer 410 of the compute fabric 400. Additionally, the software defined networking layer 410 may provide other services 415 that manage resources and/or groupings of hardware and/or software resources of the software defined networking layer 410 and/or the physical layer 405 of the compute fabric 400 that are allocated and/or utilized to support the security features of the NGPCAS 100.
As previously discussed, communications between nodes, configured containers, locations, devices, and/or other portions of the NGPCAS 100 as well as human-operated computing devices 155 (if any) may be secured via one or more VPNs, which may include mutually-exclusive and/or nested VPNs. As previously indicated, for ease of discussion and not limitations purposes, the term “VPN” is utilized herein to generally and categorically refer to various types of secured, encrypted point-to-point (PTP), peer-to-peer (P2P), and/or point-to-multipoint (PTM) connections. At any rate, each VPN, nested or otherwise, may block traffic from nodes, components, etc. that are not included in the VPN, and endpoints of the VPN may communicate with each other over the VPN via respective sessions. The VPN configuration (e.g., the number of VPNs, types of VPNs, nesting of VPNs, etc.) of the NGPCAS 100 may be implemented over one or more public and/or private networks, including private enterprise networks, the public Internet, etc. Additionally, the VPN configuration of the NGPCAS 100 may be customized to meet the security needs and requirements of the enterprise and/or of the architecture provider manager, for example. If desired, at least one of the VPNs of the NGPCAS 100 may be a permanent VPN.
Referring to
As previously mentioned, the users 155 of the NGPCAS 100 may include a human user operating a computing device at which one or more applications (e.g., a web browser, a thin client, or another user interface) execute to communicate with the NGPCAS 100, such as an operator, a configuration engineer, a third-party person who has been approved by the enterprise to access at least a portion of the system 100, another agent of the enterprise, or an agent of the architecture provider/manager 161, to name a few. Additionally or alternatively, the users 155 may include an automated user such as an external application or service that does not have a user interface and that executes on an external (e.g., remote) computing device or system. The external application/service users 155 may be enterprise-based, e.g., applications and/or services that have been configured by enterprise personnel, or may be third-party applications and/or services. Some of the external application/service users 155 may be applications and/or services provided by the architecture provider/manager 161, for utilization by the architecture provider/manager 161 and/or by enterprises. At any rate, to secure the system 100 against possible cyber-attacks when legitimate users 155 access system data and/or functionalities, each user 155 may be required to utilize one or more exposed APIs 160 to interface with and/or access the system data and/or functionalities provided by the NGPCAS 100. For example, the functionalities that are provided by the NGPCAS 100 for users 155 to utilize may be implemented in the compute fabric 102 as respective containerized components/MEEEs/granules 140, and such containerized components may be exposed to users 155 only via APIs 160 (e.g., as websites, services, etc.), where the APIs 160 may be accessed by the users 155, for example, via web browsers or thin clients. Typically, any functionality (e.g., all functionalities) that are provided by the NGPCAS 100 for users 155 to utilize is accessible to users 155 only via one or more respective APIs 160. Further, for still additional security, communications between the user 155 and the one or more APIs 160 may be secured via respective VPNs 158. For yet further security, the user 155 may be required to first be authenticated to the VPN 158 before being able to take advantage of the APIs 160, and human users in particular may be subject to multi-factor authentication in order to obtain access to the VPN 158.
In some situations, particular users 155 may be authenticated and/or authorized to utilize only a subset of the available, exposed APIs 160. That is, the set of APIs 160 that are exposed to different users 155 and/or that different users are authorized to access may differ based on the users' respective credentials. For example, authorization of different users 155 to different APIs 160 may be based on the containerized component(s) 140 (e.g., applications and/or services) to which access is provided by the APIs 160. Additionally or alternatively, authorization of different users 155 to different be APIs 160 may be implemented on an enterprise-basis (e.g., users 155 of enterprise A are allowed access to a first subset of the APIs 160 and users 155 of enterprise B are allowed access to a second, different subset of the APIs 160); on a location-basis; on a node-basis, on a user-credential basis (e.g., roles, responsibilities, and/or skills of the user), on a time-basis, and/or based on combinations thereof.
As such, by using VPNs to secure communications within the NGPCAS 100, security mechanisms which are utilized between layers of Purdue model-based systems (e.g., firewalls, data diodes, DMZs, and other mechanisms) to secure cross-level communications may be eliminated. Indeed, in an embodiment, the NGPCAS 100 does not include (e.g., excludes) any firewalls, data relays, data diodes, and DMZs that are utilized to secure communications and data delivery to, from, and within the NGPCAS 100, thereby simplifying the design, engineering, configuring, maintenance, and run-time performance of the NGPCAS 100 as compared to Purdue model-based systems. Further, as each VPN blocks or does not process any traffic that originates from outside of the VPN, and as any and all human users and/or automated users 155 of a VPN must be authenticated to the VPN, data utilized within the process control or automation system is exposed only to those components/entities that have been authorized to access the VPN over which the data is delivered. Accordingly, opportunities for externally-initiated cyber-security breaches and the spread of malware are significantly decreased as compared to present-day systems, or even, in some cases, are eliminated.
Still further, as access to selected functionalities provided by the NGPCAS 100 is provided to users 155 via APIs 160, and as users 155 must be authenticated to a VPN 158 and optionally authenticated to utilize particular APIs 160 as discussed above, cyber-security risk is even more significantly decreased as compared to the cyber-security risks of present-day systems. For example, such security techniques utilized within the NGPCAS 100 eliminate the need for any NGPCAS-specific software to be installed on some external computing devices (such as those operated by human users). That is, such external computing devices may have no (e.g., zero) installed, NGPCAS-specific software, thereby eliminating another possible avenue for cyber-security breaches. In another example, computing devices which are operated by users 155 (human or automated) may be authenticated to a VPN 158 that is not utilized by any component of the NGPCAS 100. That is, the VPN(s) that are utilized by users 155 (and optionally by the APIs 160 which are exposed to the users 155) and the VPN(s) that are utilized by other, non-user components and/or entities of the NPGCAS 100 may be mutually exclusive VPNs, thereby further eliminating other possible avenues for cyber-security breaches. In yet another example, unauthorized (but otherwise valid) users 155 may be prevented from any access at all (including read-only access) to the NGPCAS 100 or portions thereof.
As previously mentioned, each component of the NGPCAS 100 (each containerized component/MEEE/granule 140, each physical component 135, each device 105, 125, 108, 128, 148, each location 115, 118, the compute fabric 145, the architecture provider/manager 161, or generally speaking, any component which may serve as an endpoint within NGPCAS 100 networks) may be uniquely identified within the NGPCAS 100 by a unique network identifier. In embodiments, the unique network identifier of a subject component is based on an identification of the subject component as defined within a configuration database of the NGPCAS 100. In a manner similar to that discussed above for the users 155 of the NGPCAS 100, each component may be authenticated and authorized based on its unique network identifier in order for the component to access one or more VPNs, to communicate with one or more nodes, configured containers, locations, devices, and/or other components, etc. Generally speaking, components of the NGPCAS 100 (e.g., all components of the NGPCAS 100) having unique network identifiers can be discovered within the NGPCAS 100 and may be required to utilize a respective certificate to be authenticated and authorized for access.
At least some of the physical devices 105, 125, 108, 128, 148 included in the NGPCAS 100 may also include a device identifier that is unique across an enterprise, such as described above with respect to FIGS. 5A_5? In these devices, an indication of the association between the device's unique device identifier and the device's unique network identifier may be stored, e.g., within the device itself, in a configuration database of the NGPCAS 100, in a network manager of the NGPCAS 100, etc.
To further secure the NGPCAS 100, all communications that are sent and received via the networks of the NGPCAS 100 (e.g., via the VPNs 130, 158, 162 between various authenticated and authorized components) may be required to be signed and encrypted. Additionally or alternatively, plaintext protocols (such as HTTP) may be forbidden or otherwise prevented from being utilized within the NGPCAS 100. For still further security in arrangements in which an architecture provider/manager 161 manages multiple NGPCASs 100 for multiple enterprises, each enterprise may have a different Certificate Authority (CA), and self-signed certificates may be forbidden or otherwise prevented from being used. Generally speaking, to maintain security within the NGPCAS 100 over time, certificates may support revocation, have modifiable key sizes (e.g., to support system growth), and may auto-refresh without any enterprise intervention.
With particular regard to securing the compute fabric architecture 400 and components thereof, various security techniques may be employed within the NGPCAS 100. For example, as mentioned above, containerized components/MEEEs/granules 140 of the compute fabric 102/400 (e.g., APIs 160, 165; services and subsystems 435, 438, 440448, 413442 at the software defined application layer 412; services and functions 415, 418, 420, 422, 425, 452, 455, 458, 460 at the software defined network layer 410; APIs 428, 432 and/or other services provided at the adaptor layer 430, etc.) may be configured for client certificate access, where the client may be, for example, another containerized component/MEEE/granule 140, a user 155, or the architecture provider/manager 161. That is, no anonymous access to containerized components/MEEEs/granules 140 may be allowed, and only certain clients may be provided access to certain containerized components/MEEEs/granules 140 (e.g., via corresponding client certificates). Further, client certificates may be automatically and frequently rotated. Additionally, in embodiments, unused features (e.g., applications, services, etc.) may be placed in a disabled (not enabled) state.
Further, containerized components/MEEEs/granules 140 may be signed and scanned regularly for known vulnerabilities. Containerized components/MEEEs/granules 140 may be required to execute or run (e.g., always run) with least privilege, and the runtime of containerized components/MEEEs/granules 140 may be required to utilize a maximum level of container isolation, e.g., by default. In some implementations, the definition of groupings of containerized components may be prevented or restricted. For example, the NGPCAS 100 may be allowed to only define groups of containerized components that are specifically related to process control system components. For instance, a group of containerized components utilized for a bioreactor maybe defined or configured as a bioreactor grouping (e.g., by using a pod or other suitable mechanism provided by the operating system 410 of the compute fabric 400) so that the bioreactor grouping of containerized components can be co-located and moved together, e.g., to execute on different nodes, clusters, segments, etc.
At the physical layer 405 of the compute fabric 400, access to different nodes Ny, different segments of hardware, different clusters C1, . . . Cn, etc. may be controlled, e.g., on a server-basis, on an API-basis, etc. Still further, static data and disks may be encrypted at rest.
In embodiments, field devices, hardware I/O devices, gateways, and other devices may include one or more forms embedded device identification (“EDID”). In the same way that a serial number indicates a specific instance of a product, and may indicate additional information about the model, options, or other data about the product, the EDID is associated with and/or indicates an individual instance of a device or product and/or may be associated with and/or indicate additional information. As will be described below, the EDID may facilitate faster and less labor-intensive commissioning of process plants, in addition to a variety of security and other benefits.
An I/O device 522 is depicted in the block diagram of
Each EDID 510, 520, 530, 540, 560 is a built-in unique identity for the device. While, in embodiments, the EDID 510, 520, 530, 540, 560 may include multiple pieces of information, in preferred embodiments, each EDID 510, 520, 530, 540, 560 is simply a unique identifier that is associated in a relevant database with other relevant information about the device. The EDIDs 510, 520, 530, 540, 560 may be embedded on their respective devices in any number of ways. For example, in embodiments, the EDID 510, 520, 530, 540, 560 is burned into a non-volatile read-only memory (e.g., an EEPROM, UV-ROM, etc.). In other embodiments, the EDID 510, 520, 530, 540, 560 is hardwired into the device at the level of the printed circuit board (PCB) by, for example, a series of shorted and/or open signals. In still other embodiments, the EDID 510, 520, 530, 540, 560 may be embedded within a chip at manufacture. In still other embodiments, the EDID 510, 520, 530, 540, 560 may be the result of a physically unclonable function that utilizes variations in the manufacturing process of one or more components to generate what is essentially a random value. Regardless of the means of embedding the EDID 510, 520, 530, 540, 560 in the respective device, the EDID 510, 520, 530, 540, 560 is generally difficult to modify and/or over-write without, at a minimum, opening the device to access the device's internal hardware.
Particularly in embodiments in which the EDID on the device includes only a unique ID, a database 580 may associate each EDID with the values indicative of the various information 565-572 as desired and as depicted in
In embodiments, upon connection to the compute fabric 102 (e.g., upon commissioning, process start-up, reboot, etc.) each hardware device must authenticate to the system. A discovery service operating on the compute fabric 102 requests and/or receives and/or discovers from each hardware device its EDID. The discovery service transmits to the EDID service 582 (which may or may not be separate from the discovery service, in embodiments) each EDID received and, querying the database, determines one or more pieces of information associated with the EDID to determine whether to validate the EDID and, by extension, the device associated with the EDID. By way of non-limiting example, the EDID service 582 may determine: whether a given EDID is associated with the owner/enterprise/customer associated with the process plant in which the device is located; whether a given EDID is associated with the geographical region in which it is presently operating; whether a given EDID is installed at a facility for which it was intended; etc. In the event that the EDID service 582 determines that the device is valid/validated, the EDID service 582 may communicate to other services (e.g., security services, certificate authority services, etc.) that the device should be allowed to operate within the system (or to operate at all). Alternatively, if the EDID service 582 determines that the device is invalid, stolen, counterfeit, off-site, in the wrong plant, in the wrong geographical region, etc., the EDID service 582 may communicate to the other services not to allow the device to operate within the system and, in some embodiments, may remotely disable the device rendering it completely inoperable.
In this manner, the EDID may be utilized to increase security by making it more likely that devices operating in the secure environment of a process plant are of a known origin before issuing security certificates that allow the devices to connect to the secure networks over which devices and services communicate across the compute fabric 102. That is, devices that are not validated will not be granted certificates by the certificate authority. The EDID may also be utilized to prevent black market sales of devices (e.g., to contravene sanctions or trade restrictions), to deter theft, to prevent reverse engineering, to prevent counterfeiting, etc., by disabling or otherwise preventing from running any device that is off-site (not in the possession of the associated customer, not at the associated plant, not in the associated geographical region, stolen, etc.)
The implementation of the EDIDs 510, 520, 530, 540, 560 may also facilitate improved commissioning of process plants. In embodiments, the information associated with each EDID may include information about the configuration of the field device and/or the options installed on the field device. In embodiments, the information associated with each EDID may include one or more device tags and/or one or more control modules associated with the field device. As such, the discovery/EDID service may be able, based solely on the EDID, to determine, for a particular process, which services should send data to and receive data from the field device, may configure I/O services for the field device, may set up appropriate secure connections between the device and other components, etc. In short, the use of EDIDs may allow process plants to be brought online to produce products in less time (and therefore with less expense).
In addition to the security features described above, the NGPCAS 100 may include one or more other security features that protect other functionalities and aspects of the system 100, at least some of which may be provided out-of-the box, and at least some of which may be customized and adjusted, e.g., by enterprise agents, by system provider agents, etc. Such security features may include, for example, role-based access control (RBAC) of various applications and human users, service principles, security policies, security audit logs (and optionally analysis and optimization of data included therein), network security groups (e.g., by application, location, human users, and/or other categories), firewalls, and access control lists (ACLs), to name a few. Further, secrets that are utilized by the system 100 (e.g., keys, certificates, etc.) may be stored in one or more secured vaults.
Still further, resources of the system 100 may be secured in a leveled manner, such as illustrated in the example resource security technique 599 illustrated in
Additional Architectural Features of NGPCAS
Due to the decentralized and highly configurable nature of the compute fabric of the NGPCAS described herein, the NGPCAS for a particular enterprise may be configured or set up by an enterprise to enable new types of data management and execution management to be performed within the compute fabric, which enables an enterprise to uniquely configure global, or inter-plant data flow and execution management in manners not possible with previous control systems. In particular, the compute fabric of an enterprise that has multiple physical plants or locations can be set up in a hub and spoke configuration in which multiple different compute fabric “hubs” may be created to support various different physical locations or plants which are connected to the hubs via communication networks which implement the communication “spokes.” Each hub may have compute resources confined to or implemented in a particular geographical or sovereign region. These regions may be, for example, continents (e.g., North or South America, Europe, Africa, etc.), countries (e.g., the United States, Russia, China, Australia, Germany, France, etc.), states or defined regions of a particular country (e.g., California, Florida, etc.) or any other geographical or geo-political region. In this case, the compute fabric hardware may be implemented in a cloud environment or at other physical locations that is physically disposed in or contained within a particular region (or a particular set of regions) to form a compute hub. Each compute hub may be connected to one or more physical locations or plants of the enterprise via the communication infrastructure described herein that forms a spoke from the compute fabric hub to the physical location. In some cases, more than one compute fabric hub may be connected to the same physical location and each such compute fabric hub may receive all or a subset of the data from that physical location. Moreover, in some cases, a compute fabric hub may connect to one or more physical locations within the same region as the hub via one or more communication spokes, and/or may connect to physical locations in one or more different regions than the hub via other communication spokes.
Importantly, this hub and spoke configuration enables data and execution management to be configured and maintained separately at each compute fabric hub 602 to enable an enterprise having physical locations in multiple different regions comply with various different laws or data governance rules within the particular regions at which the physical locations and/or the compute fabric hubs 602 are disposed. For example, different regions (such as the United States and the EU) may have different data privacy, data export and data management laws and regulations, and so it may important to separate and track the different data that is sent to and stored at a particular compute fabric hub 602 for treatment and handling in a manner that is in compliance with appropriate laws and regulations of the hub 602. However, these laws and regulations typically apply only to data at rest, and not to data in motion. The hub and spoke structure described herein therefore also enables data (either all of the data or some subset of the data) collected by devices at a particular physical location to governed by a set of data privacy laws associated with one particular region by enabling that collected data to be sent to and only stored at a compute fabric hub 602 located in a region at which the data privacy laws and regulations are to be applied. Thus, data collected at a physical location 604A disposed in the EU may be sent directly to a compute fabric hub 602B, for example, in the United States, without being stored at the hub 602C at the EU. In this case, a direct communication spoke 603A may be established between the hub 602B in the United States and the physical location 604A in the EU and this data may or may not be sent to or stored in the hub 602C in the EU. However, in another case, the data from the physical location 604B in the EU may be first sent via a spoke 603B to the hub 602C in the EU. However, the compute fabric hub 602C in the EU may immediately send that data via the inter-hub communication spoke 6031 to the hub 602B in the United States without storing that data, to thereby assure that the data is not governed by the data laws and regulations in the EU.
Of course, the hub and spoke configuration described herein may also or instead be used to manage or direct execution and operation activities at various different hubs and/or at various different physical locations using the same concepts. For example, different compute fabric hubs 602 may manage the execution of applications and services and provide for or manage user or application authorizations differently by storing and applying different sets of rules or policies to be implemented by each compute hub 602 at the physical locations 604 to which the hubs 602 are connected. The ability of each compute fabric hub 602 to be able to store and to apply different data governance, application and other system execution rules provides for great flexibility within the enterprise in managing and storing data and in managing and controlling application execution differently at different physical locations 604 and at different compute fabric hubs 602. This feature therefore enables different configuration paradigms to be used at each different hub 602 or even at each different physical location 604, even though each of the different hubs 602 or physical locations 604 are associated with the same enterprise.
Generally speaking, to perform these data and execution management activities, the compute fabric for a particular hub 602 stores a set of rules, such as data governance rules, execution rules, access authorization rules, etc. (illustrated as components 610A, 610B, 610C, 610D and 610E at the hubs 602A, 602B, 602C, 602D and 602E, respectively), which are then implemented or applied automatically by the appropriate compute fabric components at the hubs 602A to 602E to manage data flow, application and services execution, user and services authorizations, etc. Moreover, the architecture provider/manager may provide an interface for each hub 602 to enable the enterprise (e.g., one or more authorized configuration engineers associated with the enterprise) to define, set up and store the data governance and execution rules 610 to be used at each of the hubs 602.
Additionally, as illustrated in
Still further, the NGPCAS of the chart 620 includes one or more platforms 650, which may be implemented as software as a service (SaaS) platforms in a cloud environment of the compute fabric (as illustrated by the column 624) to implement and support the application frameworks 640 and applications 630. The platforms 650 may include, for example, Kubernetes (or other) container management and orchestration platforms, observability and platform monitoring platforms (used by one or both of the enterprise or the architecture provider/manager), identity and access management platforms, security management platforms, CI/CD pipeline platforms, feature promotion platforms, compliance and governance rule storage and implementation platforms (such as the rules 610 described with respect to
Still further, as illustrated in
As will be understood, the hierarchical structure of
Configuration and Support Features of NGPCAS
As will be understood, the NGPCAS as described herein provides for or enables the software implemented components thereof to be highly configurable, transportable and editable, as most of the control and support components (e.g., control modules, containers, etc.) are located in and are executed in the compute fabric without needing to be tied to specific or predetermined computer hardware (e.g., specific servers, processors, computer nodes, etc.) This feature enables system set up and configuration activities to be performed more quickly and easily than traditional control systems, as it enables enterprise system owners or managers to store configuration components for their systems in the compute fabric, to access these components from anywhere, to copy these components to create or add additional control system structure associated with, for example, new plants or new physical locations being added to the enterprise, to new hardware installed at an existing physical location, etc. without needing to specify the location of or specifics of the computer hardware used to implement the additional configuration components. Moreover, this architecture enables an architecture provider/manager (also referred to herein as an “APM”) to simultaneously oversee the operation of multiple different enterprise systems during operation of those different enterprise systems, which enables the APM to provide both general and specific support for the different enterprise systems. For example, the APM may monitor and capture quality of service statistics and/or other data related to or defining the operation of the various software and hardware components operating in the compute fabric of the various different enterprise systems while these enterprise systems are executing to perform control and manufacturing activities. The APM can provide these measures to the different enterprise systems and can upgrade or change the configuration or usage of the computer equipment within the compute fabric, such as by adding additional compute fabric computer equipment (nodes, servers, computers, processors, etc.), by reducing compute fabric equipment (nodes, servers, computers, processors, etc.), by changing compute fabric equipment (such as using processors with higher processing speeds, larger memories, etc.) or by taking other actions in the compute fabric based on the quality of service metrics, to assure better quality of service or to reduce costs where the quality of service meets expectations (e.g., meets the quality of service metrics promised or guaranteed with a particular enterprise license).
The APM can also provide or implement various data analytic applications on the various different enterprise systems to suggest changes to those systems, which may improve operation of those systems. Still further, the APM may aggregate data from different enterprises and analyze that data to detect trends, problems, etc. and then provide general guidance to particular enterprise systems based on that analysis. The APM may, for example, use data analytics to compare the operation of control systems used for the same or similar products or manufacturing steps from the same or from different enterprise systems to determine which systems or which system configurations run better or worse than each other, which operate better or worse than an average or a baseline system, etc. The APM may then use other data analytics to determine why a particular system is running better or worse, for example, to determine if the variance is related to the presence of different equipment, different control routines, different interactions or configuration set-ups of virtual control components or actual control system hardware, etc. The APM may then produce general guidance or best practices for one or more enterprise systems based on the knowledge determined in these analyses.
Still further, architecture described herein enables faster development of and testing of components to be added to a control system as the architecture enables control or other system containers or products to be developed and provided in a container registry or in a product registry for download and implementation by an enterprise system at the will of and the timing of the enterprise system or the enterprise system manager. Feedback from the operation of these downloaded and implemented containers or products can also be automatically provided back, from the compute fabric of an enterprise, to the developer to test, upgrade and change the container or product as part of the development cycle. This architecture enables or results in a quicker development cycle as it provides for quicker implementation of new features or products and provides for automatic feedback regarding the operation of new or changed components. However, this development is still performed and implemented in a manner that enables each enterprise operator or manager to control when new containers or products are downloaded to and implemented in their systems.
Likewise, as illustrated in
Still further, any other number of enterprise (as illustrated by the dots in
As illustrated in
As will be understood, the APM 710 has a direct and secure connection into the operating networks and compute fabrics of each of the enterprises 702, 704706, etc. that the APM 710 manages or supports. As such, the APM 710 can simultaneously control the allocation of computer facilities or resources for each of the compute fabrics of any of the supported enterprises. Additionally, the APM 710 may store and execute software or data analytics modules 782 that analyze the operating data, such as metadata, from each of the compute fabrics 720, 740, 760 to calculate or determine various quality of service measurements or statistics for each of the compute fabrics 720, 740 and 760 of the enterprises 702, 704 and 706. In particular, the data analytics modules 782 may determine communication latency, CPU usage, and other computing operational statistics illustrating or defining a quality of service being provided or obtained within any of the compute fabrics 720, 740 and 760 by the computer hardware used in the compute fabrics 720, 740, 760 (regardless of who provides that computer hardware or where that computer hardware is physically located). Such data analytics or quality of service measurements may be used by the APM 710 to make changes to the underlining configuration of compute fabrics as provided by or managed by the APM 710 in order to increase or change the quality of service being provided to an enterprise or to a portion of an enterprise to, for example, meet expected, guaranteed or licensed quality standards, such as those mandated by the agreements between the APM 710 and the various enterprise owners of the enterprises 702, 704 and 706. The APM 710 may make these changes by making configuration changes within the computer hardware of the compute fabrics 720, 740, 760 via the secured connections 780 and/or may interface with any third-party provider of computer hardware or computational facilities used in the compute fabrics 720, 740 and 760, such as with a cloud computing provider like Microsoft Azure. Of course, the APM 710 may change the configuration of, the amount of, the identification of or any other configuration component of computer equipment used in or licensed from a third-party to provide the compute fabrics 720, 740760 in any desired manner. In this manner, the APM 710 has ongoing control of the quality of service and of the configuration of the computer equipment provided by or licensed by the APM 710 and provided to the enterprises 702, 704, 706 which enables the APM 710 to maintain an adequate or expected quality of service for the control systems used in or implemented by those enterprises.
Likewise, APM 710 may interface directly with user interfaces associated with the different enterprises 702, 704, 706, such as the user interfaces 730, 756 and 772, to enable enterprise managers at the user interfaces 730, 756, 772 to obtain additional licenses for additional compute fabric equipment, to spin up additional compute fabric equipment, and to provide additional or new software products or containers (as developed by the APM 710 or by a third-party developer) to the enterprise owner or manager.
In any event, the APM 710 may analyze the data it receives from the compute fabrics 720, 740, 760 in any desired manner and at any grouping level. For example, as illustrated by the chart 790 in
The APM 710 may additionally perform other types of analytics on any groupings of data associated with one or more of the enterprises to which the APM 710 is connected and may provide suggestions to the enterprises about potential changes to be made within the enterprises to increase performance, quality etc. within the enterprise or some component of the enterprise (such as at a physical location, a set of control systems, a control loop, a group of control loops of similar function, etc.) For example, the APM 710 may analyze the operation of one or more groupings of hardware, software, control systems, control loops, containers, etc. implemented within one or more enterprises to determine changes that may be made to improve control within the enterprise(s). As a particular example, the APM 710 may analyze data related to the operation of a control loop or a set of control loops within an enterprise, such as control loops used to control specific hardware at one or more plant locations, and may analyze timing signals, process variable measurements, response times, control loop statistics, etc. to determine if one or more changes to the analyzed components might provide for better performance in some manner. The APM 710 may use the results of the analyses to suggest various hardware and/or software configuration changes to be made, operations that may be performed, or actions that may be taken (such as running a tuning procedure), etc. to provide for better control operation, for example, to better control signal timing, variability, quality of product, cost of operations, etc. In some cases, the APM 710 may suggest new or different types of control or control loop algorithms, new tunings of process control devices or control loops, additional control algorithms or different control algorithms that might be useful, different equipment that might be used, different control hardware or software configurations, such as changing pinning or assigning of containers or groups of containers executed in the compute fabric, etc. Again, the analysis may be performed at any level, such as at an enterprise level, a physical plant or location level, a control loop level, a control module level, a container level, a container group level, a computer device level, etc. Of course, the APM 710 may perform other types of data analytics using data from multiple different enterprises or control systems, from a single enterprise, from a subset of components within an enterprise etc. In some cases, the APM 710 (which has access to data from multiple different enterprises) may look for commonalities or differences between operations at the different locations or different hardware of the same enterprise or at different locations or hardware of different enterprises to look for commonalities or differences in performance. The APM 710 may then perform further data analytics to determine the source or cause of those differences including differences which make the control system (control loop, control plant, etc.) operate better or worse than each other or better or worse than a baseline or average. The APM 710 may then provide the results of the analyses to the enterprises to make changes. In particular, the APM 710 may provide reports or analysis back to the enterprises to enable the enterprises to consider making changes to the control systems being implemented by the enterprises. In some cases, the APM 710 may provide one or more products or containers to be used by the enterprise to implement the suggested changes.
Additionally, as illustrated in
As will be seen,
As illustrated in
As an example, one or more of the configuration applications 802 of the configuration system 800 may be used to view the current configuration of various different elements of the enterprise, including both hardware and software elements. In addition, the one or more configuration applications 802 may be used to enable a user to make changes to the configuration of the enterprise, such as to add new hardware and/or software elements, to change one or more hardware or software elements, to delete one or more hardware or software elements, to change the manner in which one more software elements, such as containers, are pinned to other software elements or to hardware elements, to add new physical locations or hardware at one or more physical locations, to change the configuration of elements at one or more physical locations, to specify or change the configuration of computer hardware within or associated with the compute fabric of the enterprise or a portion of the enterprise, and/or to implement any other desired configuration changes.
In one example, as illustrated in more detail in
As illustrated in
Still further, the hierarchy 834 may include indications of control logic or control elements (e.g., containers) in the compute fabric under the Compute Fabric section 845. Elements in the compute fabric may include, for example, logical or virtual controllers, control modules, containers, etc., and may indicate the manner in which these elements are pinned or otherwise associated with or grouped with (e.g., assigned to) one another during runtime. The control elements may also include or illustrate digital twins associated with iOS devices or field devices within the various locations, and may include any groupings or configured groupings of containers or other elements. Likewise, other groupings of containers or control elements in the compute fabric may be listed, such as Assigned I/O, and third-party containers. Likewise, the configuration hierarchy 834 may illustrate supporting elements, such as one or more batch or continuous historians, a batch executive, recipes, advanced control elements, data analytic programs or software (such as artificial intelligence (AI) programs or algorithms), monitoring software, etc. that is tied into and operating in the enterprise, such as in the compute fabric of the enterprise. Of course, it will be understood, that a user can drill down into each of the sections of the hierarchy 834 to see or access more information and specifics about those elements and the sub-elements listed therein.
The diagraming or programming area 836 of the display 822 may be used to add, change, delete, reconfigure, program and/or otherwise create configuration elements to be installed in the enterprise, such as in the compute fabric of the enterprise or in devices at one or more of the physical locations of the enterprise. In particular, a user may select and view one or more elements of the hierarchy 834 and place these elements (or copies thereof) in the programming area 836. The user may then make changes to these elements graphically to indicate the changes to the configuration of these elements to be made. In other cases, the user may add or copy library elements within the hierarchy section 834 to the programming area 836 and may then edit those elements to create new configuration elements for a control system within the enterprise. In still other cases, a user may download or obtain one or more new configuration elements from an exterior source or database, such as from the product/container registry 807. Of course, the configuration application 802 may enable the user to add, change, delete or otherwise modify configuration elements in any other manner as well.
As a further example, the user interface 822 may provide a pop-up window 850 which may display various actions that can be taken by the configuration engineer or user in the diagram or programing area 836 to perform configuration activities. In particular, the window 850 may include a duplicate button 852, an add button 854, and upgrade button 856, an assign button 858, an import new hardware button 860, a change button 862, a deploy button 864, an implement button 866, etc. While the window 850 of
In any event, a user, such as a configuration engineer for an enterprise, may access or select some of the elements in the enterprise hierarchy 834, such as library elements or actual configuration elements, and may display or copy those elements into the configuration screen area 836 using, for example, the duplicate button 852. The user or configuration engineer may then edit these elements, which may be, for example, a control module, and entire control system, a container, a group of containers, etc., and may assign these edited or new elements to the compute fabric, to one or more new or existing devices at an already installed physical location, at one or more new physical locations, to one or more virtual controllers in the compute fabric, etc., using for example, the assign feature 858. Of course the configuration engineer may tie or assign these elements to devices (e.g., field devices) or other hardware within the enterprise. As another example, a user may use the configuration programming area 836 and/or the library 834 to assign or pin control elements or configuration elements to one another or to specific hardware or to the same hardware or to the same virtual elements (e.g., controllers) in the compute fabric of the enterprise. Of course, the user may drag and drop new or changed configuration elements within the hierarchy section 834 to assign containers such as control modules to specific virtual or physical elements.
In one example, a control routine or module 880 and a set of containers 882 are illustrated as being placed into the programming area 836. Here, a user may select particular elements in the control loop 880 or within the containers 882 and make changes to them as needed (via pop-up windows, drop down menus, etc.) to create a different control loop for a new element or new hardware, to change the configuration of an existing control loop, to change where to assign the control loop 880 or one or more of the containers 882 within the compute fabric, etc. When finished with the edits, the user may use one of the buttons in the window 850 to assign or deploy the configuration elements 880, 882 or to implement them in the compute fabric.
Still further, as illustrated in
As another example, a user may want to add a new physical location to the enterprise. In this case, the user may select a new hardware button 860 and create a new physical location in the configuration system by filling out various fields of a pop-up window. Here, the user may specify one or more hardware IDs for the new hardware, such as a gateway ID, and the configuration system or engine 804 may access that gateway and then perform an auto detect for hardware at the gateway. The configuration system may the auto-populate the hardware specifics for the new physical location in the configuration system 800. In some cases, the user may select a duplicate button 852 to duplicate hardware at another physical location for the new physical location where the new physical location is designed to use the same basic hardware or field devices to perform a process as an existing physical location. The addition of a new physical location is illustrated by the dotted lines 740D in
In another example, the user may simply want to add additional hardware at one of the already establish locations such as the location 740C of
Still further, the user may use other buttons such as the change button 862 in order to reassign or change the assignment of one or more elements in the computer fabric in order to assign these elements to particular hardware, to particular locations, etc. For example, as illustrated
Of course, the user may take any other actions via the configuration system of
Of course, a new product or container placed into the registry could be any configuration element or elements and may be, for example, new or updated control elements, display elements, communication elements, data analytic elements, etc. As will be understood, the configuration element development at the boxes or steps 902 and 904 are performed, in a general sense, off-line from operation of an enterprise that is to use that element until the product or container is ready for deployment in the enterprise. Of course, the fact that a new or changed product or container (which may be an upgrade to an existing product or container) is available within the registry 910, 920 may be pushed out to the enterprise in the form of marketing or other information provided to the enterprise or to a user at the enterprise to enable the enterprise to know that the new container or product exists. In some cases, the new containers or products may be upgrades to existing containers or products already being used by the enterprise or the containers or products may be new containers or products that may be need a new license or an additional license by the enterprise in order to be downloaded and used at the enterprise.
In any event, once one of the registries 910 and/or 920 store a new or enhanced product or container, the enterprise owner may then access and download those containers or products at their leisure or convenience, without needing to have these products pushed to the enterprise at times when the enterprise is not ready to absorb them or use them. Thus, using this system, the enterprise owner 906 has full control of when it implements or installs new or upgraded products or containers. When ready, the enterprise 906 (or a user at the enterprise 906) may access and download a new or upgraded product or container from one of the registries 910, 920, such as using the system of
Importantly, the product development cycle 900 provides an easy and quick development cycle that enables rapid deployment of new products (e.g., containers) being created for an enterprise system 906. However, this development cycle also enables the enterprise owner to download and install or deploy new configuration elements at the convenience of and on the time schedule of the enterprise, instead of on the time schedule of the product developer as is common today. Still further, while only one enterprise 906 is illustrated in the loop or configuration development cycle 900 of
Enterprise-Level and Provider Features Provided by the NGPCAS
The Next Generation Process Control and Automation System (NGPCAS) described in the present disclosure may be used to support various enterprise-level services and functionalities, as well as services and functionalities of the provider of the NGPCAS with respect to enterprises. Enterprise-level services and functionalities, generally speaking, may execute to support implementation of one or more industrial and/or automation processes by each enterprise respectively. Each enterprise may, respectively, implement one or more industrial and/or automation processes, with each process being implemented among one, two, three, four or more physical locations or sites. At a very high level, and as will be described in further detail herein, the various enterprise-level services and functionalities may support the execution of processes themselves implemented by the enterprise (e.g., system configuration, control, operations, maintenance, data storage and management, etc.), support the execution of the underlying services serving as the functional building blocks of the compute fabric (e.g., creating and relocating containerized services), and support the management of the NGPCAS architecture and service/application framework by the system provider of the NGPCAS architecture (e.g., through compute fabric monitoring, physical layer resource allocation and monitoring, containerized service management, application distribution and upgrade management, etc.). The enterprise-level services and functionalities (also referred to herein as “applications”) can be executed at least partially (and, in some cases, entirely) in the compute fabric of the NGPCAS architecture of the present disclosure (e.g., the NGPCAS architecture as described with respect to the foregoing figures, for example with services/applications herein being implemented as software modules in the application layer of the NGPCAS compute fabric). Provider services and functionalities, generally speaking, may execute to support operation of NGPCASs (and processes involved therein) for one, two, three, four or more enterprises.
The compute fabric, while executing any one or more services/applications on behalf of an enterprise or the provider, may be communicatively connected with the pool of physical process equipment in one or more physical locations via one or more transport networks as described in the present disclosure. The transport networks(s) may, for example, include one or more point-to-point, peer-to-peer, and/or point-to-multipoint connections (such as VPNs) to securely connect any containerized software component to another containerized component and/or a specific physical component to exchange information as needed by the service(s)/application(s) (e.g., real-time and/or historical process data, physical site or process configuration information, site personnel information, etc.). A human user may interact with the service(s)/application(s) via one or more client computing devices (e.g., laptop, tablet, mobile computing devices, etc.), using a secure communicative connection with the portion of the compute fabric implementing the service(s)/applications. The communicative connection may, for example, utilize a VPN, and the execution of the service(s)/application(s) at the client device may include executing a particular containerized component of the compute fabric that has been configured and authorized to run the service/application at the client device. Thus, without further repetition of the description of the NGPCAS as provided in the foregoing sections of this disclosure, it should be appreciated that execution of the services/applications described in this section may utilize any suitable ones of the structures, features, and functionalities of the NGPCAS previously described in this disclosure (e.g., containerized services, digital twins, VPNs and/or other communications/security features, user identity/access security features, process hardware security features, etc.).
Real-Time Control, Operations, and Monitoring Features
Services/applications supported by the NGPCAS may, at a high level, include various service/applications that support real-time control, operations, and/or monitoring of one or more physical sites operated by the enterprise (each physical site having physical devices of the enterprise), one or more NGPCASs of the enterprise, or of an entire enterprise (e.g., including multiple physical sites and/or multiple NGPCASs). While some of these services/applications support automatic control/operations/monitoring of the physical site(s) (e.g., without intervention by a human user), various services/functionalities described herein may further include exposing selected information to one or more human users (e.g., via dashboards) and/or receiving commands from the one or more human users to further support the control/operations/monitoring of the physical site(s). The term “enterprise user” will be used herein to refer to a human user in an enterprise that has purchased NGPCAS resources/capabilities from a system provider of the NGPCAS. “Provider” or “manager,” as used herein, may refer to a human user which may have exclusive access to certain services or applications associated with management/administration of the NGPCASs of a plurality of enterprises. It should be appreciated that access of these services/functionalities by the enterprise user or provider may be subject to various authorization/authentication controls for the enterprise user or provider and for the client computing device(s) that the enterprise user or provider utilizes to access the services/functionalities. Generally speaking, whereas an enterprise user may generally have access to services/functionalities associated with the enterprise of which the user is a part, a provider/manager user may have access to services or applications associated with many NGPCASs or enterprises. Enterprise users and provider/manager users may have different permissions to access, view, and/or manipulate data at different times, with each of the enterprise user and provider/manager user permissions being respectively defined based upon policies that may be defined by the enterprise, by the system provider, or by some combination thereof.
The client device(s) may access these services/functionalities via the compute fabric from any location, e.g., from within a physical site to which the services/functionalities pertain, from within another physical site operated by the enterprise, from still another physical location agnostic to any particular physical site (e.g., a home office or a remote service center servicing one, two, three or more sites simultaneously), or otherwise from various remote locations (e.g., a mobile device).
Compute-fabric-based control, operations, and maintenance services/functionalities envisioned herein include, but are not necessarily limited to:
Turning first to
Turning to
The enterprise explorer region 1032 further includes a template library explorer 1033 that allows the user to navigate between template objects, each of which may be usable to configure and arrange new instances of sites, process modules, field devices, control loops, I/O networks, etc. For example, by interacting with a bioreactor template shown in the library explorer 1033, the user may summon a “Create Unit Wizard” region 1034 enabling the user to define, from the preconfigured bioreactor template, a bioreactor to be configured for one or more of the physical sites of the enterprise as defined in the region 1032. The template for the bioreactor may, for example, define bioreactor properties such as process material inputs/outputs, operating parameters of the bioreactor, etc. Continuing from
Containerized Service Management Features
As discussed throughout the present disclosure, any of the services/functionalities of the NGPCAS described herein may be implemented as one or more sets of configured containers (“containerized services”) which execute in the compute fabric to perform the actions attributed thereto (e.g., as described with respect to the compute fabric 400 in
First, the containerized services executing any particular application (e.g., control containerized services) may be executed anywhere, e.g. at any physical location and/or in the remote compute fabric. Moreover, regardless of where the containerized services execute (e.g., on- or off-premises), the applications described herein may, in embodiments, remain accessible to users located at any physical location. For example, one or more applications operating a particular physical site (e.g., executing a control loop, performing site monitoring, etc.) may be implemented by a portion of the compute fabric in a location separate from the particular physical site (e.g., at another physical site, or at a remote operations facility which may jointly operate one, two, three, four or more sites). As another example, an application executing enterprise-wide functionalities pertaining to multiple physical sites of the enterprise (e.g., an enterprise-wide monitoring application) may be executed via portions of the compute fabric physically located on-premises at a particular site, at combinations of two or more sites, and/or at a site-agnostic remote location. In any case, for any remotely-executing containerized services operating on behalf of a particular site, compute-fabric resources on-premises at the site may execute redundant containerized services, such that control of the process plant can fail over to the on-premises compute fabric if necessary (e.g., in the event of failure of resources in the remote compute fabric, or loss of remote communicative connection to the site). Fail-overs may, for example, include (1) a fail-over from a node at a particular physical site to another node in the particular physical site of the enterprise, (2) a fail-over from a node at a particular physical site to a different physical site, (3) a fail-over from a node in the remote compute fabric to a node in a different portion of the remote compute fabric, or (4) a fail-over from a node in the remote compute fabric to a node in a physical site of the enterprise.
The containerized services executing any particular application may also be instantiated from any physical location and/or in the remote compute fabric. In other words, the location at which a containerized service is instantiated (manually by a human user, and/or automatically via a portion of the compute fabric) may be the same as, or different from, the location at which the instantiated containerized service ultimately executes. In some embodiments, for any containerized service instantiated within the NGPCAS, an orchestration service (e.g., orchestration service 422 from
Moreover, the NGPCAS supports the movement of the execution of any particular containerized service(s) across two or more locations (e.g., via the Orchestration service 422 from
In some implementations, execution of containerized services of an enterprise using the NGPCAS is moved regionally or globally across two, three, four or more physical locations to locate the execution of containerized services proximate to monitoring/control/operation facilities that are active and on-line in the NGPCAS at any given time. For example, execution of containerized services may be moved multiple times per-day to “follow the sun” across multiple remote locations, such that execution of the services is performed at each of the multiple locations during daylight hours of the respective locations (e.g., for an enterprise with remote monitoring/control/operation facilities in or near the cities of Denver, Frankfurt, and Sydney, execution of the services could be moved daily across portions of the compute fabric near the respective facilities (e.g., in the physical sites themselves or in remote locations nearer to such physical sites), such that services execute near each respective location when the respective location is between 09:00 and 17:00 hours). Thus, execution of services is moved dynamically to place the services close to personnel who can provide daytime service to the NGPCAS at any given hour. Of course, in alternate embodiments, execution of the containerized services need not be moved in order to control remote facilities. That is, in embodiments, one, two, three, or more enterprise facilities may operate to control multiple plant sites. For example, three control facilities may each operate in respective locales such that each control facility operates during daytime hours in its respective locale and such that, during those hours, the operating control facility coordinates control of all active plant facilities for the enterprise. Doing so can reduce overtime costs and, additionally, can provide high personnel redundancy with fewer additional personnel. In any event, in the case that containerized services are moved, the NGPCAS may perform automatic tuning of one or more control algorithms of one or more physical sites of the enterprise to account for differences in communication latencies that may be introduced by the movement of the containerized services.
In view of the above-described capabilities of the NGPCAS with regard to the containerized services, various application services/functionalities may be envisioned to allow enterprise users (or, in some instances, provider/manager users) to view and/or modify the locations and/or portions of the compute fabric via which other applications execute.
For example, in some implementations, any of the control, operations, and/or monitoring functionalities for a particular physical site may further provide an indication of where the underlying containerized services for the application are presently executing, and/or where redundant containerized services for the same application are located (e.g., in the physical site being controlled/operated/monitored, in a different physical site, or in any particular one of a number of remote operations centers). In another example implementation, a unified dashboard application provided by the NGPCAS for an enterprise may provide a unified view of containerized service execution locations for each of a plurality of services/applications in the enterprise NGPCAS and/or for a plurality of sites operated by the enterprise via the NGPCAS.
In any case, the NGPCAS may provide functionalities to enable to the user to cause transfer of the execution of containerized services for any given service/functionality between two locations (e.g., on-premises to off-premises, or vice versa). The user may cause the transfer of execution of the services to be performed immediately, performed at a future scheduled time (e.g., to “follow the sun” by moving the services at shift change between two or more disparate operations centers), and/or conditionally based upon performance measurements at the location from which the service execution is to be moved (e.g., containerized services are moved when the present execution location experiences a falls below a threshold network bandwidth, falls below a threshold processing availability measurement, experiences greater than a threshold latency time in essential communications with a physical site, etc.). Additionally or alternatively, in the event that a configured container fails and falls offline, the dashboard application may display an indication of failure of the configured container and enable the user to restore a state of operation of the configured container from a backup of the configured container (e.g., an on-premises backup).
Modular/Subscription-Based Distribution Features for Process-Related Services/Applications
In view of the wide variety of industrial/automation processes that may be implemented by enterprises NGPCASs, the utility of various NGPCAS services/applications may vary significantly. Advantageously, the NGPCAS architecture of the present description allows an enterprise to acquire (e.g., purchase or subscribe to) and utilize these services/applications “a la carte,” (i.e., independently of other services/applications). In other words, an enterprise user can individually select particular control, operations, monitoring, analytics, diagnostics, and/or other services/applications to implement in the enterprise NGPCAS (or a particular portion of the NGPCAS, such as a particular sites) to serve the particular needs of the enterprise.
In an implementation, a system provider hosts a marketplace or “store” that enables an enterprise-level user to acquire (e.g., purchase or subscribe to) various services/applications published by the system provider (e.g., as described with respect to “Example 3” in the foregoing sections of the present disclosure). The system provider may, for example, publish various control services/applications/modules that are particularly suitable for certain processes or process modules (e.g., for batch processes, for continuous processes, for processing of a particular process material through a particular process equipment, etc.). As another example, the system provider may publish various analysis, monitoring and/or diagnostics applications that are suitable for various process operations. As still another example, the system provider may publish still other system configuration tools, resource management tools, and/or other tools not specific to process operations but still pertaining to an enterprise's ownership or use of an NGPCAS. Applications may be packaged or distributed in various manners, e.g., some applications may be pre-packaged together as a unit, or some applications may be packaged together for inclusion upon initial purchase of NGPCAS resources by an enterprise.
In implementations, third-parties can similarly provide services/applications for purchase/subscription by other enterprises, e.g., for the other enterprise to access from a third-party via one or more APIs. A first enterprise, for example, may generate and distribute process functionalities utilized by the first enterprise and allow one or more other enterprises to run their own instances of the same functionalities via the respective NGPCASs of the other one or more enterprises of their own respective uses (e.g., one or more other enterprises may have interest in a control tool, analytics tool, or diagnostics tool offered by the first enterprise that implements a same or similar industrial process).
Upon an enterprise user acquiring a service/application from the marketplace (e.g., via subscription or via one-time-purchase), the system provider may automatically instantiate the acquired service/application into the compute fabric to enable the enterprise to utilize the acquired service/application (e.g., as described in this disclosure). In the case of services/applications that may materially affect operation of a physical site of the enterprise (e.g., an acquired control/operation algorithm, alarm module, etc.), the enterprise NGPCAS may first implement the acquired service/application in parallel with active control elements (e.g., in “simulation mode”) to determine whether the service/application will operate safely and/or whether the service/application will improve operation of the physical site of the enterprise (e.g., based upon simulated process variable measurements, process setpoints, product quality metrics, process environmental metrics, etc.).
In view of the above, various NGPCAS marketplace/store functionalities may be envisioned. Any of these functionalities may be accessed and utilized by the authenticated/authorized user(s) from a suitable computing device(s) from any location using the compute fabric, e.g., from within a physical site to which the services/functionalities pertain, from within another physical site, or from still another physical location agnostic to any particular physical site. Moreover, the marketplace/store functionalities may enable an enterprise NGPCAS to utilize obtained services/applications in any portion of the NGPCAS once the enterprise has obtained the services/applications. By way of example and not limitation, NGPCAS marketplace/store functionalities may include any one or more of the following:
In view of the foregoing,
First looking to
Centralized Monitoring and NGPCAS Upgrades by System Provider
Still additionally to the capabilities the NGPCAS described above, the system provider of the NGPCAS may implement further system monitoring and administrative management functionalities with respect to respective NGPCASs operated by each of one, two, three, four or more enterprises (independently from the enterprises themselves). Generally speaking, centralized monitoring (and/or optimization) functionalities described herein are provided not only for process control functionalities (e.g., operations directly participating in implementation of a process), but also for quality of service and resource management of NGPCAS resources across an entire enterprise. In various implementations, enterprise users may customize the usage (or non-usage) of these multi-enterprise monitoring functionalities by the NGPCAS, e.g., by selecting preferences and permissions for data to be shared with the system provider for purposes of multi-enterprise monitoring (e.g., sharing of site-specific process data, compute fabric network traffic, application usage by enterprise users, etc.).
In some implementations, for example, the system provider may monitor process data (e.g., process setpoints, control inputs/outputs, measured process variables, product quality metrics, process environment data, site-specific network traffic in controllers, field devices, etc.) in one or more physical sites of the enterprise to identify and notify an enterprise user of potentially unsafe conditions in one or more sites (e.g., field device failures, unsafe environmental conditions, unsafe product quality, etc.). As another example, the system provider may monitor higher-level network traffic in the enterprise NGPCAS (e.g., across the compute fabric servicing one or more sites) to identify anomalies in use of the NGPCAS network by processes and/or users (e.g., potentially indicative of intrusions into a process control system, or compromise of a user account of an enterprise user). In some embodiments, the monitoring of process data and/or network traffic of the enterprise includes comparing the process data/network traffic to baselines established by comparison to one or more other enterprise NGPCASs (e.g., other enterprises of similar size, running similar processes at physical sites, etc.). Moreover, the monitoring of the NGPCAS may implement fault-handling and/or fault-protection measures for the enterprise, e.g., to protect against a large scale failure in one or more of the enterprise's physical sites or in large portions of the remote compute fabric (for example, the system provider may manage reallocation of the serving compute fabric among various remote compute fabric resources, or delegate primary operations for the enterprise to on-premises portions of the compute fabric).
In addition to performing centralized monitoring of multiple enterprises, the system provider can centrally manage hardware and software upgrades for the respective NGPCASs of a plurality of enterprises. For example, as described in foregoing sections of this disclosure, the system provider may centrally receive and implement upgrades to services/applications provisioned via the system provider marketplace (e.g., upgrades to first-party and/or third-party services/applications, implemented via new, upgraded containerized services created in the NGPCASs of enterprises that have acquired the service/application). Upgrades managed by the service provider can additionally or alternatively include control algorithms, field device or controller software, diagnostics software, etc. Moreover, as an enterprise installs new hardware (e.g., controllers, field devices, etc.), in one or more physical sites, the discovery service operating in the compute fabric may automatically discover the new hardware (e.g., based upon respective EDID detected from the hardware upon startup). Upon detecting the new hardware, the compute fabric may automatically upgrade and/or replace compute fabric resources of the enterprise to match the new hardware in the physical site (e.g., such that compute fabric representations of the hardware, including the digital twin of the hardware, match the hardware actually installed in the physical site).
Although the monitoring functionalities and graphical user interfaces described above are described as being accessed and utilized by the system provider, it should be appreciated that at least some of these monitoring functionalities/interfaces can be provided to any particular enterprise. In such instances, monitoring functionalities and information provided to an enterprise user are limited to the activity of that particular enterprise (e.g., the enterprise's own process data, network traffic, compute fabric usage, etc., without compromising information associated with other enterprises served by the system provider). Centralized management/monitoring by a particular enterprise may, for example, include monitoring of one, two, three, four or more physical sites of the enterprise, from any physical location (e.g., from within a site or from one or more locations agnostic to any particular site). Centralized management/monitoring by the particular enterprise can, for example, include monitoring operation of physical processes at one or more sites and/or monitoring of NGPCASs resources of the enterprise (e.g., on- or off-premises NGPCAS resources). Additionally or alternatively, an enterprise user may centrally manage various system upgrades and/or optimizations within one or more NGPCASs of the enterprise or within components of one or more NGPCASs of the enterprise (e.g., upgrades/optimizations to process control, operations, monitoring, etc., as described above with respect to the system provider).
Other Considerations
When implemented in software, any of the applications, modules, etc. described herein may be stored in any tangible, non-transitory computer readable memory such as on a magnetic disk, a laser disk, solid state memory device, molecular memory storage device, or other storage medium, in a RAM or ROM of a computer or processor, etc. Although the example systems disclosed herein are disclosed as including, among other components, software and/or firmware executed on hardware, it should be noted that such systems are merely illustrative and should not be considered as limiting. For example, it is contemplated that any or all of these hardware, software, and firmware components could be embodied exclusively in hardware, exclusively in software, or in any combination of hardware and software. Accordingly, while the example systems described herein are described as being implemented in software executed on a processor of one or more computer devices, persons of ordinary skill in the art will readily appreciate that the examples provided are not the only way to implement such systems.
Thus, while the present application has been described with reference to specific examples, which are intended to be illustrative only and not to be limiting, it will be apparent to those of ordinary skill in the art that changes, additions or deletions may be made to the disclosed embodiments without departing from the spirit and scope of the application.
The particular features, structures, and/or characteristics of any specific embodiment may be combined in any suitable manner and/or in any suitable combination with one and/or more other embodiments, including the use of selected features with or without corresponding use of other features. In addition, many modifications may be made to adapt a particular application, situation and/or material to the essential scope or spirit of the present application. It is to be understood that other variations and/or modifications of the embodiments of the present application described and/or illustrated herein are possible in light of the teachings herein and should be considered part of the spirit or scope of the present application. Certain aspects of the application are described herein as example aspects.
This application claims priority as a continuation of U.S. patent application Ser. No. 18/223,384, entitled “Process Control and Automation System Architecture,” which was filed on Jul. 18, 2023, which in turn claims priority under 35 U.S.C. § 119 to each of U.S. Provisional Patent Application Ser. No. 63/390,238, entitled “Next Generation Process Control and Automation System,” which was filed Jul. 18, 2022; U.S. Provisional Patent Application Ser. No. 63/398,441, entitled “Securing Next Generation Process Control and Automation Systems,” which was filed Aug. 16, 2022; U.S. Provisional Patent Application Ser. No. 63/417,861, entitled “Configuration Features of Next Generation Process Control and Automation Systems,” which was filed Oct. 20, 2022; and U.S. Provisional Patent Application Ser. No. 63/418,006, entitled “Enterprise-Level Features Provided by the NGPCAS,” which was filed Oct. 20, 2022, the entire disclosures of each of which are hereby expressly incorporated by reference herein.
| Number | Date | Country | |
|---|---|---|---|
| 63418006 | Oct 2022 | US | |
| 63417861 | Oct 2022 | US | |
| 63398441 | Aug 2022 | US | |
| 63390238 | Jul 2022 | US |
| Number | Date | Country | |
|---|---|---|---|
| Parent | 18223384 | Jul 2023 | US |
| Child | 18370691 | US |
