Efficient region selection system for executing request-driven code

Abstract

An on-demand code execution environment present in points of presence (POPs) and in regions serviced by the POPs is provided herein. For example, a POP may receive a request to execute a task associated with user-defined code. If the POP determines that the computing resources necessary to execute a received task are not available or that the POP should not execute the received task for another reason (e.g., the task is not commonly received and the computing resources needed to execute the task are therefore best allocated for other requests), the POP can identify a region suitable for executing the task and forward the task to the identified region. An on-demand code execution environment present in the identified region can execute the task and forward the results of the execution to the POP for distribution back to a user device that requested the task execution.

Description

BACKGROUND

Computing devices can utilize communication networks to exchange data. Companies and organizations operate computer networks that interconnect a number of computing devices to support operations or to provide services to third parties. The computing systems can be located in a single geographic location or located in multiple, distinct geographic locations (e.g., interconnected via private or public communication networks). Specifically, data centers or data processing centers, herein generally referred to as a “data center,” may include a number of interconnected computing systems to provide computing resources to users of the data center. The data centers may be private data centers operated on behalf of an organization or public data centers operated on behalf, or for the benefit of, the general public.

To facilitate increased utilization of data center resources, virtualization technologies allow a single physical computing device to host one or more instances of virtual machines that appear and operate as independent computing devices to users of a data center. With virtualization, the single physical computing device can create, maintain, delete, or otherwise manage virtual machines in a dynamic manner. In turn, users can request computer resources from a data center, including single computing devices or a configuration of networked computing devices, and be provided with varying numbers of virtual machine resources.

In some scenarios, virtual machine instances may be configured according to a number of virtual machine instance types to provide specific functionality. For example, various computing devices may be associated with different combinations of operating systems or operating system configurations, virtualized hardware resources and software applications to enable a computing device to provide different desired functionalities, or to provide similar functionalities more efficiently. These virtual machine instance type configurations are often contained within a device image, which includes static data containing the software (e.g., the OS and applications together with their configuration and data files, etc.) that the virtual machine will run once started. The device image is typically stored on the disk used to create or initialize the instance. Thus, a computing device may process the device image in order to implement the desired software configuration.

BRIEF DESCRIPTION OF DRAWINGS

Throughout the drawings, reference numbers may be re-used to indicate correspondence between referenced elements. The drawings are provided to illustrate example embodiments described herein and are not intended to limit the scope of the disclosure.

FIG. 1 is a block diagram of an illustrative operation environment in which a plurality of POPs may implement an on-demand code execution environment and a plurality of regional data centers may implement an on-demand code execution environment.

FIG. 2A illustrates an example block diagram of the auxiliary services 122 of FIG. 1, according to one embodiment.

FIG. 2B illustrates an example block diagram of the on-demand code execution environments and of FIG. 1, according to one embodiment.

FIG. 3A is a block diagram of the operating environment of FIG. 1 illustrating the operations performed by the components of the operating environment to execute a task, according to one embodiment.

FIG. 3B is another block diagram of the operating environment of FIG. 1 illustrating the operations performed by the components of the operating environment to execute a task, according to one embodiment.

FIG. 4 is a block diagram of the operating environment of FIG. 1 illustrating the operations performed by the components of the operating environment to replicate user-defined code to one or more geographic regions, according to one embodiment.

FIG. 5 is an example table stored in the replication data store of FIG. 1, according to one embodiment.

FIG. 6 is a flow diagram depicting a task execution routine illustratively implemented by a POP, according to one embodiment.

FIG. 7 is another flow diagram depicting a task execution routine illustratively implemented by a POP, according to one embodiment.

DETAILED DESCRIPTION

Generally described, aspects of the present disclosure relate to executing user-defined code within a low latency, on-demand code execution environment, as well as managing the computing devices within the code execution environment on which the code is executed. The on-demand code execution environment may operate as part of a system of rapidly provisioned and released computing resources, often referred to as a “cloud computing environment.” Specifically, the code execution environment may include one or more computing devices, virtual or non-virtual, that are “pre-warmed” (e.g., booted into an operating system and executing a complete or substantially complete runtime environment) and configured to enable execution of user-defined code, such that the code may be executed rapidly without initializing the virtual machine instance. Each set of code on the on-demand code execution environment may define a “task,” and implement specific functionality corresponding to that task when executed on the on-demand code execution environment. Individual implementations of the task on the on-demand code execution environment may be referred to as an “execution” of the task. By defining tasks on the on-demand code execution environment and selectively executing those tasks, users may implement complex functionalities at high speed and low latency, without being required to deploy, configure, or manage the computing devices on which the tasks are executed. The on-demand code execution environment, in turn, may execute tasks of multiple users simultaneously, thus allowing efficient use of computing resources of those devices. To ensure the security and privacy of user information, the on-demand code execution environment may generally ensure that tasks of each user are executed on distinct computing devices (which may be virtual computing devices), thus reducing the chances that a task executing on behalf of a first user could interfere with or gain information regarding execution of a task on behalf of a second user.

In some instances, an on-demand code execution environment may operate as a distributed system in which multiple points of presence (POPs) implement instances of the on-demand code execution environment. As used herein, a POP is intended to refer to any collection of related computing devices utilized to implement functionality on behalf of one or many providers. POPs are generally associated with a specific geographic location in which the computing devices implementing the POP are located, or with a region serviced by the POP. For example, a data center or a collection of computing devices within a data center may form a POP. An on-demand code execution environment may utilize multiple POPs that are geographically diverse, to enable users in a variety of geographic locations to quickly transmit and receive information from the on-demand code execution environment. In some instances, the POPs may also implement other services, such as content delivery network (CDN) services, data storage services, data processing services, etc. For the purposes of the present disclosure, these other services will generally be referred to herein as “auxiliary services.” Implementation of auxiliary services and instances of the on-demand code execution environment on the same POP may be beneficial, for example, to enable tasks executed on the on-demand code execution environment to quickly interact with the auxiliary services.

However, implementation of auxiliary services on a POP may also limit the amount of computing resources available to implement an instance of the on-demand code execution environment. For example, a POP implementing an edge server of a CDN service may have relatively little available computing resources (e.g., in the form of disk space, processor time, central processing power, graphical processing power, memory, network bandwidth, internal bus utilization, etc.) with which to execute tasks. These computing resources may be even further depleted by attempting to execute those tasks within a distinct computing device, such as a virtual computing device, that does not implement functionalities of the CDN service. Moreover, the available physical space to house the edge server may be limited, thereby limiting the amount of computing resources that can be added to the POP to account for the relatively little available computing resources. In addition, dedicating a virtual or non-virtual computing device to a single user can negatively impact the performance of the on-demand code execution environment by reducing the flexibility of the environment in allocating computing resources for the execution of tasks requested by other users. Thus, if a POP receives a sufficient number of requests to execute tasks, the POP may run out of computing resource capacity to execute the tasks. Once the POP no longer has computing resource capacity to handle the execution of a task, the POP may return an exception to the requesting entity indicating that no more computing resource capacity is available to execute the task.

Accordingly, aspects of the present disclosure enable the on-demand code execution environment to be present in POPs and in regions serviced by the POPs such that tasks can be executed regardless of whether there is a sufficient amount of computing resources available at the POP to handle task requests. Thus, if a POP determines that the computing resources necessary to execute a received task are not available (e.g., the POP lacks computing resource capacity) or that the POP should not execute the received task for another reason (e.g., the task is not commonly received and the computing resources needed to execute the task are therefore best allocated for other requests), the POP can forward the task to a region that the POP services for execution by an on-demand code execution environment present in the region. For example, if the POP implements an edge server of a CDN service, the edge server may receive from a user device a request to execute a task. The edge server may then determine whether to instruct an on-demand code execution environment local to the POP to execute the task. Some factors that the edge server may consider in making the determination include how busy the POP is (e.g., the amount of unused computing resources currently available), the popularity of the requested task (e.g., how often the POP receives requests to perform the task relative to other requested tasks), a time it may take to execute the task locally as opposed to within a region, the historical volume of requests received from the user device, the time of day that the request is received, the latency-sensitivity of the task (e.g., based on an analysis of the user-defined code), properties of the task (e.g., whether execution of the user-defined code causes the retrieval of content and, if so, whether such content is available in a local POP cache or at an origin server), and/or the like.

If the edge server determines to instruct the local on-demand code execution environment to execute the task, then the edge server forwards the requested task to the local on-demand code execution environment for execution. As described in greater detail below, the local on-demand code execution environment may instruct an existing virtual machine instance to execute the requested task or may provision a new virtual machine instance and instruct the new virtual machine instance to execute the requested task. Once the execution is complete, the local on-demand code execution environment may forward the execution results to the edge server for transmission back to the user device.

If, on the other hand, the edge server determines not to instruct the local on-demand code execution environment to execute the task, then the edge server forwards the requested task to a server in a region. The edge server may forward the task to a server in the region that is closest geographically to the POP. Alternatively, the edge server may identify or select one region from a set of regions to receive the task. Some factors that the edge server may consider in making the determination include whether a region has an updated version of the user-defined code used to execute the task, whether a region has previously received a request to execute the task and/or if the region currently has a virtual machine instance provisioned to execute such tasks, and/or the like. Once the region is selected, then the edge server forwards the task to a server in the selected region.

Upon receiving the task from the edge server, the server in the region can instruct the on-demand code execution environment local to the region to execute the task. Similar to as described above, the on-demand code execution environment in the region may instruct an existing virtual machine instance to execute the requested task or may provision a new virtual machine instance and instruct the new virtual machine instance to execute the requested task. Once the execution is complete, the on-demand code execution environment in the region can forward the execution results to the region server. The region server can then transmit the execution results to the edge server in the POP for distribution to the user device.

Generally, user-defined code is specific to a region and stored therein. Thus, to enable multiple regions to execute tasks corresponding to user-defined code associated with a particular region, each region may include a replication system configured to replicate user-defined code associated with the respective region to other regions. The replication system may track which regions to replicate the user-defined code to and the status of such replications. The replication system may periodically forward this information to the POP to aid the edge server in determining which region to forward a task request.

As used herein, the term “virtual machine instance” is intended to refer to an execution of software or other executable code that emulates hardware to provide an environment or platform on which software may execute (an “execution environment”). Virtual machine instances are generally executed by hardware devices, which may differ from the physical hardware emulated by the virtual machine instance. For example, a virtual machine may emulate a first type of processor and memory while being executed on a second type of processor and memory. Thus, virtual machines can be utilized to execute software intended for a first execution environment (e.g., a first operating system) on a physical device that is executing a second execution environment (e.g., a second operating system). In some instances, hardware emulated by a virtual machine instance may be the same or similar to hardware of an underlying device. For example, a device with a first type of processor may implement a plurality of virtual machine instances, each emulating an instance of that first type of processor. Thus, virtual machine instances can be used to divide a device into a number of logical sub-devices (each referred to as a “virtual machine instance”). While virtual machine instances can generally provide a level of abstraction away from the hardware of an underlying physical device, this abstraction is not required. For example, assume a device implements a plurality of virtual machine instances, each of which emulates hardware identical to that provided by the device. Under such a scenario, each virtual machine instance may allow a software application to execute code on the underlying hardware without translation, while maintaining a logical separation between software applications running on other virtual machine instances. This process, which is generally referred to as “native execution,” may be utilized to increase the speed or performance of virtual machine instances. Other techniques that allow direct utilization of underlying hardware, such as hardware pass-through techniques, may be used, as well

While a virtual machine executing an operating system is described herein as one example of an execution environment, other execution environments are also possible. For example, tasks or other processes may be executed within a software “container,” which provides a runtime environment without itself providing virtualization of hardware. Containers may be implemented within virtual machines to provide additional security, or may be run outside of a virtual machine instance.

To execute tasks, the on-demand code execution environment described herein may maintain a pool of pre-initialized virtual machine instances that are ready for use as soon as a user request is received. Due to the pre-initialized nature of these virtual machines, delay (sometimes referred to as latency) associated with executing the user-defined code (e.g., instance and language runtime startup time) can be significantly reduced, often to sub-100 millisecond levels. Illustratively, the on-demand code execution environment may maintain a pool of virtual machine instances on one or more physical computing devices, where each virtual machine instance has one or more software components (e.g., operating systems, language runtimes, libraries, etc.) loaded thereon. When the on-demand code execution environment receives a request to execute the program code of a user, which specifies one or more computing constraints for executing the program code of the user, the on-demand code execution environment may select a virtual machine instance for executing the program code of the user based on the one or more computing constraints specified by the request and cause the program code of the user to be executed on the selected virtual machine instance. The program codes can be executed in isolated containers that are created on the virtual machine instances. Since the virtual machine instances in the pool have already been booted and loaded with particular operating systems and language runtimes by the time the requests are received, the delay associated with finding compute capacity that can handle the requests (e.g., by executing the user code in one or more containers created on the virtual machine instances) is significantly reduced.

The on-demand code execution environment may include a virtual machine instance manager configured to receive user code (threads, programs, etc., composed in any of a variety of programming languages) and execute the code in a highly scalable, low latency manner, without requiring user configuration of a virtual machine instance. Specifically, the virtual machine instance manager can, prior to receiving the user code and prior to receiving any information from a user regarding any particular virtual machine instance configuration, create and configure virtual machine instances according to a predetermined set of configurations, each corresponding to any one or more of a variety of run-time environments. Thereafter, the virtual machine instance manager receives user-initiated requests to execute code, and identify a pre-configured virtual machine instance to execute the code based on configuration information associated with the request. The virtual machine instance manager can further allocate the identified virtual machine instance to execute the user's code at least partly by creating and configuring containers inside the allocated virtual machine instance. Various embodiments for implementing a virtual machine instance manager and executing user code on virtual machine instances is described in more detail in U.S. patent application Ser. No. 14/502,648, entitled “PROGRAMMATIC EVENT DETECTION AND MESSAGE GENERATION FOR REQUESTS TO EXECUTE PROGRAM CODE” and filed Sep. 30, 2014 (hereinafter referred to as “the '648 application”), the entirety of which is hereby incorporated by reference herein. Additional details regarding the on-demand code execution environment are provided in U.S. patent application Ser. No. 14/971,859, entitled “EXECUTION LOCATIONS FOR REQUEST-DRIVEN CODE” and filed Dec. 16, 2015 (hereinafter referred to as “the '859 application”), the entirety of which is hereby incorporated by reference herein.

The foregoing aspects and many of the attendant advantages of this disclosure will become more readily appreciated as the same become better understood by reference to the following detailed description, when taken in conjunction with the accompanying drawings.

Example Multi-Region On-Demand Code Execution Environment

FIG. 1 is a block diagram of an illustrative operating environment 100 in which a plurality of POPs 120 may implement an on-demand code execution environment 124 and a plurality of regional data centers 130 may implement an on-demand code execution environment 132. The POPs 120 may further implement auxiliary services 122 and a user and task history data store 126. Various user devices 102 may communicate with the POPs 120 via a network 110 to request the execution of tasks. Tasks may be written, by way of non-limiting example, in JavaScript (e.g., node.js), Java, Python, and/or Ruby (and/or another programming language). Requests to execute a task may generally be referred to as “calls” to that task. Such calls may include the user-defined code (or the location thereof) to be executed and one or more arguments to be used for executing the user-defined code. For example, a call may provide the user-defined code of a task along with the request to execute the task. In another example, a call may identify a previously uploaded task by its name or an identifier. In yet another example, code corresponding to a task may be included in a call for the task, as well as being uploaded in a separate location (e.g., storage of auxiliary services 122 or a storage system internal to the on-demand code execution environments 124 and/or 132) prior to the request being received by the on-demand code execution environments 124 and/or 132. The on-demand code execution environments 124 and/or 132 may vary its execution strategy for a task based on where the code of the task is available at the time a call for the task is processed.

While the user devices 102 and POPs 120 are shown as grouped within FIG. 1, the user devices 102 and POPs 120 may be geographically distant, and independently owned or operated. For example, the user devices 102 could represent a multitude of users in various global, continental, or regional locations accessing the POPs 120. Further, the POPs 120 may be globally, continentally, or regionally disparate, in order to provide a wide geographical presence for the on-demand code execution environment 124 and/or the auxiliary services 122. Accordingly, the groupings of user devices 102 and POPs 120 within FIG. 1 is intended to represent a logical, rather than physical, grouping. The POPs 120 of FIG. 1 are illustratively shown as implementing both auxiliary services 122 and instances of the on-demand code execution environment 124. However, the operating environment 100 may additionally or alternatively include POPs that execute only auxiliary services 122 or only an instance of the on-demand code execution environment 124. Components of the auxiliary services 122 are described in greater detail below with respect to FIG. 2A and components of the on-demand code execution environments 124 and 132 are described in greater detail below with respect to FIG. 2B.

As illustrated in FIG. 1, each regional data center 130 may include a region server 131, an on-demand code execution environment 132, a replication system 133, a cache data store 134, a replication data store 135, and a code usage data store 136. The components of the regional data center 130 may be capable of storing user-defined code, replicating user-defined code to other regional data centers 130, and executing tasks. The regional data centers 130 may each be globally, continentally, or regionally disparate. For example, each regional data center 130 may located within a geographical region serviced by a POP 120. While FIG. 1 illustrates each POP 120 communicating with a single regional data center 130, this is not meant to be limiting. A single POP 120 may communicate with one or more regional data centers 130 (e.g., regional data centers 130 that are within a threshold distance of the location of the POP 120).

The operating environment 100 further includes one or more origin servers 104. The origin servers 104 may include any computing device owned or operated by an entity that has provided one or more sets of content (“distributions”) to a CDN for subsequent transmission to user devices 102. For example, origin servers 104 may include servers hosting web sites, streaming audio, video, or multimedia services, data analytics services, or other network-accessible services. The origin servers 104 may include primary versions of content within various distributions. If the POPs 120 function as CDNs, the primary versions of content may be retrieved by the various POPs 120 for subsequent transmission to the user devices 102. In an embodiment, the POPs 120 includes a cache that stores frequently-requested content (e.g., service data store 208) and the regional data centers 130 include caches that store frequently-requested content (e.g., the cache data store 134). If requested content is not present in the POP 120 cache, the POP 120 may first request the content from the regional data center 130. If the requested content is also not present in the cache data store 134, then the POP 120 may retrieve the content from an origin server 104. In addition, origin servers 104 may be included in a regional data center 130, in addition to or as an alternative to cache data store 134.

Users, by way of user devices 102, may interact with the on-demand code execution environments 124 and/or 132 to provide executable code, and establish rules or logic defining when and how such code should be executed on the on-demand code execution environments 124 and/or 132. For example, a user may wish to run a piece of code in connection with a web or mobile application that the user has developed. One way of running the code would be to acquire virtual machine instances from service providers who provide infrastructure as a service, configure the virtual machine instances to suit the user's needs, and use the configured virtual machine instances to run the code. In order to avoid the complexity of this process, the user may alternatively provide the code to the on-demand code execution environments 124 and/or 132, and request that the on-demand code execution environments 124 and/or 132 execute the code using one or more pre-established virtual machine instances. The on-demand code execution environments 124 and 132 can handle the acquisition and configuration of compute capacity (e.g., containers, instances, etc., which are described in greater detail below) based on the code execution request, and execute the code using the compute capacity. The on-demand code execution environments 124 and/or 132 may automatically scale up and down based on the volume, thereby relieving the user from the burden of having to worry about over-utilization (e.g., acquiring too little computing resources and suffering performance issues) or under-utilization (e.g., acquiring more computing resources than necessary to run the codes, and thus overpaying).

In an embodiment, a user device 102 may transmit a request to execute a task to POP 120A or 120B via the network 110. The request may be received by a server in the auxiliary services 122 (e.g., server 207). The server may then determine whether to execute the task locally or to forward the task to a region server 131 in one of the regional data centers 130. For example, the server may use information stored in the user and task history data store 126 as well as information identifying how busy the POP 120 is (e.g., the amount of unused computing resources currently available to the POP 120), a time it may take to execute the task locally at the POP 120 (e.g., which is based on properties of the user-defined code used to execute the task and the available computing resources, such as whether a virtual machine instance is already provisioned to execute the task), a time it may take to execute the task within a regional data center 130 (e.g., which is based on properties of the user-defined code used to execute the task and whether a virtual machine instance is already provisioned in the on-demand code execution environment 132 to execute the task), the time of day that the request is received, and/or the like to determine whether to execute the task locally or to forward the task to a region server 131.

The information stored in the user and task history data store 126 may include the popularity of the requested task (e.g., how often the POP 120 receives requests to perform the task relative to other requested tasks), the historical volume of requests received from the user device 102 that requested the task execution, the latency-sensitivity of the task, properties of the task (e.g., what type of content is retrieved as a result of execution of the task), and/or the like. The latency-sensitivity of the task and/or the properties of the task may be determined based on a previous or current analysis of the user-defined code. As an example, the user-defined code may be stored within a regional data center 130 that is associated with the same region as the user-defined code. The server in the auxiliary service 122 may retrieve the user-defined code from the appropriate regional data center 130 (e.g., at a previous time so that the POP 120 could execute a previously-received task or at a current time) and analyze the user-defined code to estimate whether a use case associated with the task is sensitive to delays in task execution and/or to identify other properties of the task. As an illustrative example, based on an analysis of the user-defined code, the server may determine that execution causes a script to be run (e.g., Javascript). Other components or code modules may be dependent on the completion of the script (e.g., other components to be loaded in order to display a network or content page) and therefore the task may be latency sensitive. On the other hand, the analysis may result in a determination that the task relates to decompressing bits, which is less latency sensitive. As another illustrative example, based on an analysis of the user-defined code, the server may determine that execution of the task results in the retrieval of content stored in an origin server 104.

The server may consider some or all of the above-described factors independently or in combination to determine whether to execute the task locally or remotely in a regional data center 130. For example, if the server determines that the amount of unused computing resources currently available to the POP 120 is not sufficient to execute the task, then the server determines to forward the task to a region server 131. Otherwise, if the server determines that the amount of unused computing resources currently available to the POP 120 is sufficient to execute the task and no other factors indicate that the task should be executed remotely, then the server may instruct the on-demand code execution environment 124 to execute the task.

As another example, if the server determines that the time it would take to execute the task locally at the POP 120 is faster than the time it would take to execute the task remotely in a regional data center 130, then the server may instruct the on-demand code execution environment 124 to execute the task. Otherwise, if the server determines that the time it would take to execute the task locally at the POP 120 is slower than the time it would take to execute the task remotely in a regional data center 130, then the server may forward the task to the regional data center 130.

As another example, if the task is not commonly received (e.g., the task has not been received before, the number of times the task has been received is below a threshold value, the number of times the task has been received as a percentage of a number of times any task has been received is below a threshold value, etc.) and/or the user device 102 typically sends a low volume of requests (e.g., the user device 102 sends a number of requests below a threshold value, the percentage of requests sent by the user device 102 as a percentage of all requests received is below a threshold value, a time required by the user device 102 to execute all desired tasks is below a threshold value, etc.), then the server may forward the task to the regional data center 130. Provisioning a virtual machine instance to service a user device 102 and/or request and tearing down the virtual machine instance when tasks are completed can be time and resource intensive, and therefore the server may determine to offload the task to a regional data center 130 if the task is uncommon or the user device 102 sends a low volume of requests (and thus does not require the virtual machine instance for a long period of time). Otherwise, if the task is commonly received (e.g., the number of times the task has been received is above a threshold value, the number of times the task has been received as a percentage of a number of times any task has been received is above a threshold value, etc.) and/or the user device 102 typically sends a high volume of requests (e.g., the user device 102 sends a number of requests above a threshold value, the percentage of requests sent by the user device 102 as a percentage of all requests received is above a threshold value, a time required by the user device 102 to execute all desired tasks is above a threshold value, etc.), then the server may instruct the on-demand code execution environment 124 to execute the task.

As another example, if the task request is received during a time of day in which the POP 120 is busy (e.g., a threshold percentage of the POP 120 computing resources are occupied) and the task is not commonly received and/or the user device 102 typically sends a low volume of requests, then the server may forward the task to the regional data center 130. Otherwise, if the task request is received during a time of day in which the POP 120 is busy and the task is commonly received and/or the user device 102 typically sends a high volume of requests and/or if the task request is receive during a time of day in which the POP 120 is not busy, then the server may instruct the on-demand code execution environment 124 to execute the task.

As another example, if the server determines that the task is latency sensitive, then the server may instruct the on-demand code execution environment 124 to execute the task. Requesting the on-demand code execution environment 124 to execute the task may be faster than the regional data center 130 because of the fewer number of transmissions required to complete execution of the task (e.g., the server may not communicate directly with the on-demand code execution environment 132, so there may be additional transmissions required to forward the task to the on-demand code execution environment 132). Otherwise, if the server determines that the task is not latency sensitive and no other factors indicate that the task should be executed locally, then the server may forward the task to the regional data center 130.

As another example, if the server determines that the user-defined code properties indicate that execution of the task results in the retrieval of content stored in an origin server 104, then the server may forward the task to the regional data center 130. As described above, the POP 120 may send a request for content to the regional data center 130 first before sending a request to the origin server 104 if the content is not available in the POP 120 cache. Thus, given the limited computing resources available at the POP 120, the server may forward the task to be executed remotely because a request may be sent to the regional data center 130 anyway. Otherwise, the server may instruct the on-demand code execution environment 124 to execute the task if no other factors indicate that the task should be executed remotely.

In an embodiment, once the server determines to forward the task to a regional data center 130, the server forwards the task to the region server 131 in the regional data center 130 that is geographically closest to the POP 120. For example, the regional data center 130A may be the closest regional data center to the POP 120A. Thus, the server in the auxiliary services 122A may forward the task to the region server 131A.

In other embodiments, once the server determines to forward the task to a regional data center 130, the server then determines which regional data center 130 to send the request to. For example, it may be more efficient to forward the task to a first regional data center 130 than a second regional data center 130 even if the first regional data center 130 is farther geographically from the POP 120 than the second regional data center 130. Such situations may occur when the first regional data center 130 has received requests to execute the task before and therefore has provisioned a virtual machine instance to execute the task, whereas the second regional data center 130 has not received requests to execute the task or has not received many requests to execute the task and the time required to provision a virtual machine instance to execute the task would be longer than the time required to send the task to the first regional data center 130 for execution. In other instances, the first regional data center 130 may store a current or updated version of the user-defined code used to execute the task, whereas the second regional data center 130 stores an old or invalid version of the user-defined code.

The user and task history data store 126 may further store data identifying which regional data centers 130 have previously received requests to execute the task and/or have virtual machine instances provisioned to execute the task and which regional data centers 130 are storing a current version of the user-defined code used to execute the requested task (or any version of the user-defined code used to execute the requested task). The server can use this information to then select a regional data center 130 to receive the forwarded task. For example, the server may narrow the selection by considering regional data centers 130 that store a current version of the user-defined code. From this filtered list of regional data centers 130, the server may determine which regional data center(s) 130 have received requests to execute the task and/or have virtual machine instances provisioned to execute the task. If more than one regional data center 130 satisfies this criteria, then the server may select the regional data center 130 from those that satisfy the criteria that is closest to the POP 120. If no regional data center 130 has received requests to execute the task and/or does not have a virtual machine instance provisioned to execute the task, then the server may select the regional data center 130 that is closest to the POP 120. If no regional data center 130 accessible by the POP 120 has a current version of the user-defined code, then the server may determine whether it is necessary for the user-defined code to be current for the task to be executed. For example, the task request may include an indication of what version of the user-defined code to use to execute the task, the minimum version number of the user-defined code that is sufficient to execute the task, and/or whether the version number matters for execution of the task. If it is not necessary for the user-defined code to be current for the task to be executed, then the server may forward the task to the closest regional data center 130 that has previously received requests to execute the task and/or a virtual machine instance provisioned to execute the task (if present) or the closest regional data center 130 if no regional data center 130 has previously received requests to execute the task and/or has a virtual machine instance provisioned to execute the task. If it is necessary for the user-defined code to be current for the task to be executed, then the server may return an exception to the user device 102 indicating that a current version of the user-defined code is not available.

If the server determines to forward the task to a regional data center 130 that is not the closest regional data center 130 (or any other regional data center 130) because the closest regional data center 130 does not have a virtual machine instance provisioned to execute the task, the server may also instruct the closest regional data center 130 (or any other regional data center 130) to provision a virtual machine instance for executing the task. The server may continue to forward future tasks to the non-closest regional data center 130 until the virtual machine instance is provisioned and ready to execute the task. Once the virtual machine instance is provisioned (e.g., once the server is notified by the regional data center 130 that the virtual machine instance is provisioned), then the server may start forwarding future tasks to the closest regional data center 130 (or any other regional data center 130 instructed to provision a virtual machine instance for executing the task).

As described above, the region server 131 receives a task forwarded by the server in the auxiliary services 122. The region server 131 may then forward the task to the on-demand code execution environment 132 for execution. After execution is complete, the on-demand code execution environment 132 forwards the execution results to the region server 131. The region server 131 can then forward the execution results to the server, which then forwards the execution results to the user device 102 that requested the task execution.

As described above, tasks are executed according to user-defined code. A user (using, for example, a user device 102) may compose code for use in executing a task. The user-defined code may be associated with a particular geographic region selected by the user and stored within the regional data center 130 associated with the selected geographic region (e.g., stored within data store 260 of the on-demand code execution environment 132). To enable multiple regional data centers 130 to execute a task, the user can identify one or more other geographic regions that are authorized to execute the task. Alternatively, a default set of geographic regions can be authorized to execute the task. The authorized geographic regions can be stored in the replication data store 135 in the regional data center 130 associated with the user-defined code's geographic region in an entry associated with the user-defined code.

Once other geographic regions are authorized to execute the task, the replication system 133 in the regional data center 130 associated with the user-defined code's geographic region may replicate the bits of the user-defined code and transmit the replicated bits to the replication systems 133 in the regional data centers 130 associated with the authorized geographic regions. The replication system 133 in the regional data center 130 associated with the user-defined code's geographic region may also replicate the bits of the user-defined code and transmit the replicated bits to the replication systems 133 in the regional data centers 130 associated with the authorized geographic regions when the user-defined code is updated. The replication system 133 that replicated the user-defined code and transmitted the replicated bits can periodically request a status of the transfer from the other replication systems 133 that are receiving the replicated bits. Statuses can include a version of the user-defined code currently stored in the regional data center 130, whether the latest replication transmission is complete (and if not, the progress of the transmission), and/or whether the replication transmission has started. The replication system 133 can store the received statuses in the replication data store 135 in an entry associated with the user-defined code. Periodically, the replication system 133 can retrieve the statuses from the replication data store 135 and transmit the statuses to the POP 120 associated with the regional data center 130 for storage in the user and task history data store 126. As described above, the server in the auxiliary services 122 can use this information to select a regional data center 130 to receive a forwarded task.

As an illustrative example, a user may define code that is associated with the same geographic region as the regional data center 130A. Thus, the user-defined code may be stored in the data store 260 of the on-demand code execution environment 132A. A user may identify a geographic region associated with the regional data center 130B as a geographic region that is authorized to execute the task and this information may be stored in the replication data store 135A. Thus, the replication system 133A may replicate the user-defined code and transmit the replicated user-defined code to the replication system 133B. The replication system 133A may periodically request a status of the transfer from the replication system 133B. The received status may also be stored in the replication data store 135A. The replication system 133A may periodically transmit the received status to the POP 120A for storage in the user and task history data store 126A. The server 207A in the auxiliary services 122A can then use the status information to aid in identifying or selecting a regional data center 130 to execute the task.

In some embodiments, the POP 120 can forward the received statuses to other POPs 120 such that some or all POPs 120 are aware of which regional data centers 130 have current user-defined code and which do not. Alternatively, the replication system 133 can forward the statuses to multiple POPs 120 to achieve the same objective.

In further embodiments, one regional data center 130 can forward a task execution request to another regional data center 130 (assuming that the other regional data center 130 is authorized to execute the task). For example, the POPs 120 may not be aware of (or may not have current information of) which regional data centers 130 have previously received task requests and/or which regional data centers 130 have a virtual machine instance provisioned to execute the task. However, the regional data centers 130 may store such information. As an example, the code usage data store 136 may store information indicating which user-defined code has been used by the respective on-demand code execution environment 132 to execute a task in the past, which regional data centers 130 have previously executed a task or provisioned a virtual machine instance to execute a task, and/or which regional data centers 130 have a current version of the user-defined code. The regional data centers 130 may communicate with each other to share such information. If a region server 131 receives a task request and the region server 131 determines that such a request has not been previously received and/or a virtual machine instance is not provisioned to execute the task, the region server 131 may query the code usage data store 136 to determine whether another regional data center 130 has previously received a request to execute the task and/or has provisioned a virtual machine instance to execute the task. If such a regional data center 130 exists, the region server 131 may forward the task to the region server 131 of that regional data center 130. The results of executing the task can be transmitted back to the regional data center 130 that originally received the forwarded task request before being transmitted to the requesting POP 120 and the user device 102.

As an illustrative example, the region server 131A may receive a task request from the POP 120A. However, the on-demand code execution environment 132A may not have a virtual machine instance provisioned to execute the task. The on-demand code execution environment 132B, though, may have a virtual machine instance provisioned to execute the task. Thus, the region server 131A may forward the request to the region server 131B for execution. The region server 131B can then forward the request to the on-demand code execution environment 132B. Results of executing the task may be forwarded by the on-demand code execution environment 132B to the region server 131B, the region server 131B may forward the results to the region server 131A, the region server 131A may forward the results to the server 207A in the auxiliary services 122A, and the server 207A may forward the results to the user device 102 that requested execution of the task.

Various example user devices 102 are shown in FIG. 1, including a desktop computer, laptop, and a mobile phone, each provided by way of illustration. In general, the user devices 102 can be any computing device such as a desktop, laptop or tablet computer, personal computer, wearable computer, server, personal digital assistant (PDA), hybrid PDA/mobile phone, mobile phone, electronic book reader, set-top box, voice command device, camera, digital media player, and the like. The on-demand code execution environments 124 and/or 132 may provide the user devices 102 with one or more user interfaces, command-line interfaces (CLI), application programing interfaces (API), and/or other programmatic interfaces for generating and uploading user-executable code, invoking the user-provided code (e.g., submitting a request to execute the user codes on the on-demand code execution environments 124 and/or 132), scheduling event-based jobs or timed jobs, tracking the user-provided code, and/or viewing other logging or monitoring information related to their requests and/or user codes. Although one or more embodiments may be described herein as using a user interface, it should be appreciated that such embodiments may, additionally or alternatively, use any CLIs, APIs, or other programmatic interfaces.

The network 110 may include any wired network, wireless network, or combination thereof. For example, the network 110 may be a personal area network, local area network, wide area network, over-the-air broadcast network (e.g., for radio or television), cable network, satellite network, cellular telephone network, or combination thereof. As a further example, the network 110 may be a publicly accessible network of linked networks, possibly operated by various distinct parties, such as the Internet. In some embodiments, the network 110 may be a private or semi-private network, such as a corporate or university intranet. The network 110 may include one or more wireless networks, such as a Global System for Mobile Communications (GSM) network, a Code Division Multiple Access (CDMA) network, a Long Term Evolution (LTE) network, or any other type of wireless network. The network 110 can use protocols and components for communicating via the Internet or any of the other aforementioned types of networks. For example, the protocols used by the network 110 may include Hypertext Transfer Protocol (HTTP), HTTP Secure (HTTPS), Message Queue Telemetry Transport (MQTT), Constrained Application Protocol (CoAP), and the like. Protocols and components for communicating via the Internet or any of the other aforementioned types of communication networks are well known to those skilled in the art and, thus, are not described in more detail herein.

FIG. 2A illustrates an example block diagram of the auxiliary services 122 of FIG. 1, according to one embodiment. As illustrated in FIG. 2A, the auxiliary services 122 can include a server 207 communicatively coupled to a service data store 208. The server 207 and service data store 208 may operate in conjunction to implement functionalities of the auxiliary services 122. For example, where the auxiliary services 122 is an edge server for a CDN, the server 207 and service data store 208 may operate to cache distributed content (e.g., as provided by a user of the auxiliary services 122) and respond to requests from end users for such cached content. As a further example, where the auxiliary services 122 is a database system, the server 207 and service data store 208 may operate to facilitate and manage interactions with a database. In general, auxiliary services 122 may include any network-based service or data source.

Auxiliary services 122 may be associated with operation of the on-demand code execution environment 124. For example, the server 207 may determine whether a requested task should be executed locally by the on-demand code execution environment 124 or remotely by a regional data center 130 in a manner as described above. If the server 207 determines that a regional data center 130 should receive the task for execution, the server 207 may further determine which regional data center 130 should receive the task in a manner as described above. In some instances, auxiliary services 122 actively transmit information, such as task requests (e.g., in the form of API calls) or other task-triggering information, to the on-demand code execution environment 124. In other instances, auxiliary services 122 may be passive, such that data is made available for access by the on-demand code execution environment 124. For example, components of the on-demand code execution environment 124 may periodically poll such passive data sources, and trigger execution of tasks within the on-demand code execution environment 124 based on the data provided (e.g., based on the availability of a task request received from a user device 102).

Operation of various auxiliary services 122, including CDN networks, database services, data storage services, and data processing services, are known within the art, and therefore will not be described herein. While a simplified view of auxiliary services 122 is shown in FIG. 2A (e.g., including a single server 207 and service data store 208), the POP 120 may implement auxiliary services 122 by use of any number of computing or storage devices, which may not be shown in FIG. 2A. In some instances, computing or storage devices associated with auxiliary services 122 may also be utilized to implement functionalities of the on-demand code execution environment 124. For example, virtual machine instances 250 of the on-demand code execution environment 124 (which are described in more detail below) may be implemented by the server 207. In some instances, the on-demand code execution environment 124 may execute tasks directly on the server 207 (e.g., without use of a virtual machine instance 250).

FIG. 2B illustrates an example block diagram of the on-demand code execution environments 124 and 132 of FIG. 1, according to one embodiment. As illustrated in FIG. 2A, the on-demand code execution environment 124 and the on-demand code execution environment 132 may each include the same components. However, the on-demand code execution environment 124 and the on-demand code execution environment 132 may receive data from different sources.

The on-demand code execution environments 124 and 132 include a frontend 220, worker manager 230, instance pool 240, and data stores 260 collectively configured to enable users (via user devices 102) or regional data centers 130 to submit computer executable instructions (also referred to herein as “code,” “program code,” or “user-defined code”) to the on-demand code execution environment 124 or 132 for execution as a “task.” For example, the data store 260 in the on-demand code execution environment 124 may store a copy of the code defined by a user. The code stored in the on-demand code execution environment 124 may be received from an on-demand code execution environment 132. The data store 260 in the on-demand code execution environment 132 may store a primary copy of the code defined by a user. The code stored in the on-demand code execution environment 132 may be received from a user device 102. Thus, the on-demand code execution environment 132 may store the primary, read-write version of the code, whereas the on-demand code execution environment 124 may store a secondary, read-only version of the code. An update to the code stored in the on-demand code execution environment 132 may be propagated (e.g., by the replication system 133) to other regional data centers 130 and/or POPs 120.

The frontend 220 can facilitate interactions between the on-demand code execution environment 124 and the user devices 102, auxiliary services 122, and/or other computing devices (not shown in FIG. 1 or 2B) via the network 110. Similarly, the frontend 220 can facilitate interactions between the on-demand code execution environment 132 and the region server 131, the user devices 102, and/or other computing devices (not shown in FIG. 1 or 2B) via the network 110. These interactions may include, for example, submission of code, which may be stored within the data stores 260, or transmission of requests to execute code, which may be communicated to the worker manager 230 for assignment to and execution by a virtual machine instance 250 within the instance pool 240. The frontend 220 may receive calls to execute tasks from the server 207 (if in the on-demand code execution environment 124) or the region server 131 (if in the on-demand code execution environment 132) in response to Hypertext Transfer Protocol Secure (HTTPS) requests (e.g., task requests) from a user. Also, any information (e.g., headers and parameters) included in the HTTPS request may also be processed and utilized when executing a task. Any other protocols, including, for example, HTTP, MQTT, and CoAP, may be used to transfer the message containing a task call to the frontend 220.

The on-demand code execution environments 124 and 132 are depicted in FIG. 2B as operating in a distributed computing environment including several computer systems that are interconnected using one or more computer networks (not shown in FIG. 1 or 2B). The on-demand code execution environments 124 and 132 could also operate within a computing environment having a fewer or greater number of devices than are illustrated in FIG. 1 or 2B. Thus, the depiction of the on-demand code execution environments 124 and 132 in FIGS. 1 and 2B should be taken as illustrative and not limiting to the present disclosure. For example, the on-demand code execution environments 124 and 132 or various constituents thereof could implement various Web services components, hosted or “cloud” computing environments, and/or peer to peer network configurations to implement at least a portion of the processes described herein. Further, the on-demand code execution environments 124 and 132 may be implemented directly in hardware or software executed by hardware devices and may, for instance, include one or more physical or virtual servers implemented on physical computer hardware configured to execute computer executable instructions for performing various features that will be described herein. The one or more servers may be geographically dispersed or geographically co-located, for instance, in one or more POPs 120 or regional data centers 130.

The frontend 220 can distribute a request to execute a task to a worker manager 230, which can assign tasks to virtual machine instances 250 for execution. In the example illustrated in FIG. 2B, the worker manager 230 manages the instance pool 240, which is a group (sometimes referred to as a pool) of virtual machine instances 250 that are utilized to execute tasks. As shown in FIG. 2B, instances 250 may have operating systems (OS) 252, language runtimes 254, and containers 256. Containers 256 are logical units created within a computing device using the resources available on that device, and may be utilized to isolate execution of a task from other processes (e.g., task executions) occurring on the device. For example, in order to service a request to execute a task, the worker manager 230 may, based on information specified in the request, create a new container 256 within a virtual machine instance 250 within which to execute the task. In other instances, the worker manager 230 may locate an existing container 256 in one of the instances 250 in the instance pool 240 and assign the container 250 to handle the execution of the task. Containers 256 may be implemented, for example, as Linux containers. The containers 256 may have individual copies of the OSes 252, the runtimes 254, and user-defined code 258 corresponding to various tasks assigned to be executed within the container 256.

While the instance pool 240 is shown in FIG. 2B as a single grouping of virtual machine instances 250, some embodiments of the present application may separate virtual machine instances 250 that are actively assigned to execute tasks from those virtual machine instances 250 that are not actively assigned to execute tasks. For example, those virtual machine instances 250 actively assigned to execute tasks may be grouped into an “active pool,” while those virtual machine instances 250 not actively assigned to execute tasks may be placed within a “warming pool.” Those virtual machine instances 250 within the warming pool may be pre-initialized with an operating system, language runtimes, or other software required to enable rapid execution of tasks in response to user requests. Further details regarding active pools and warming pools are described in greater detail within the '648 application, which is incorporated by reference above (e.g., at FIG. 1 of the '648 application).

On receiving a request to execute a task, the worker manager 230 may locate a virtual machine instance 250 within the instance pool 240 that has available capacity to execute the task. The worker manager 230 may further create a container 256 within the virtual machine instance 250, and provision the container 256 with any software required to execute the task (e.g., an operating system 252, runtime 254, or code 258). For example, a container 256 is shown in FIG. 2B provisioned with operating system 252B, runtime 254B, and a set of code 258. The operating system 252B and runtime 254B may be the same as the operating system 252A and runtime 254A utilized by the virtual machine instance 250, or may be different. After provisioning the container 256 with the requisite software for a task, the worker manager 230 can cause the virtual machine instance 256 to execute the task on behalf of a user.

The worker manager 230 includes a processing unit, a network interface, a computer readable medium drive, and an input/output device interface, all of which may communicate with one another by way of a communication bus. The network interface may provide connectivity to one or more networks or computing systems. The processing unit may thus receive information and instructions from other computing systems or services via the network 110. The processing unit may also communicate to and from a memory of the worker manager 230 and further provide output information for an optional display via the input/output device interface. The input/output device interface may also accept input from an optional input device. The memory may contain computer program instructions (grouped as modules in some embodiments) that the processing unit executes in order to implement one or more aspects of the present disclosure.

Additional details of the components and functionality of the auxiliary services 122 and the on-demand code execution environments 124 and 132 are described in greater detail in the '859 application, which is incorporated by reference above.

Example Block Diagrams for Executing a Task

FIG. 3A is a block diagram of the operating environment 100 of FIG. 1 illustrating the operations performed by the components of the operating environment 100 to execute a task, according to one embodiment. As illustrated in FIG. 3A, the user device 102 calls a task by transmitting a request to execute the task to the server 207 at (1). The server 207 may query the user and task history data store 126 for task data at (2). The task data may include the popularity of the requested task, the historical volume of requests received from the user device 102 that requested the task execution, the latency-sensitivity of the task, properties of the task, and/or the like. The server 207 may then use the task data, and information identifying how busy the POP 120 is, a time it may take to execute the task locally at the POP 120, a time it may take to execute the task within a regional data center 130, the time of day that the request is received, and/or the like to determine whether to execute the task locally at (3). For example, the server 207 may determine whether sufficient computing resources are available to execute the task locally (and may determine to execute the task locally if the POP 120 has sufficient computing resources available). As another example, the server 207 may determine whether requests to execute the task are commonly received (and may determine to execute the task locally if requests to execute the task are common). If the server 207 determines to execute the task locally, then the server 207 may instruct the on-demand code execution environment 124 to execute the task at (4). Otherwise, if the server 207 determines not to execute the task locally, the server 207 may forward the task to the region server 131 at (7).

As described above, the server 207 may forward the task to the region server 131 even if the POP 120 has sufficient computing resource capacity to execute the task. For example, the server 207 may use information stored in the user and task history data store 126 as well as information identifying how busy the POP 120 is, a time it may take to execute the task locally at the POP 120, a time it may take to execute the task within a regional data center 130, the time of day that the request is received, and/or the like to determine whether to execute the task locally or to forward the task to the region server 131.

If the server 207 instructs the on-demand code execution environment 124 to execute the task, then the on-demand code execution environment 124 executes the task at (5). After execution, the on-demand code execution environment 124 reports results of the execution to the server 207 at (6).

If the server 207 forwards the task to the region server 131, the region server 131 then instructs the on-demand code execution environment 132 to execute the task at (8). In response to the instruction, the on-demand code execution environment 132 executes the task at (9) and reports results of the execution to the region server 131 at (10). The region server 131 then reports the execution results to the server 207 at (11).

Once the server 207 receives the execution results (either from the on-demand code execution environment 124 or from the region server 131), then the server 207 reports the execution results to the user device 102 at (12). Thus, the task may be successfully executed regardless of whether the POP 120 in which the server 207 is present has sufficient computing resource capacity to execute the task.

FIG. 3B is another block diagram of the operating environment 100 of FIG. 1 illustrating the operations performed by the components of the operating environment 100 to execute a task, according to one embodiment. As illustrated in FIG. 3B, the user device 102 calls a task by transmitting a request to execute the task to the server 207 at (1). The server 207 may query the user and task history data store 126 for task data at (2). The task data may include the popularity of the requested task, the historical volume of requests received from the user device 102 that requested the task execution, the latency-sensitivity of the task, properties of the task, and/or the like. The server 207 may then use the task data, and information identifying how busy the POP 120 is, a time it may take to execute the task locally at the POP 120, a time it may take to execute the task within a regional data center 130, the time of day that the request is received, and/or the like to determine whether to execute the task locally at (3). For example, the server 207 may determine whether sufficient computing resources are available to execute the task locally (and may determine to execute the task locally if the POP 120 has sufficient computing resources available). As another example, the server 207 may determine whether requests to execute the task are commonly received (and may determine to execute the task locally if requests to execute the task are common). If the server 207 determines to execute the task locally, then the server 207 may instruct the on-demand code execution environment 124 to execute the task at (4). Otherwise, if the server 207 determines not to execute the task locally, the server 207 may select a region server 131 in a particular geographic region to execute the task.

As described above, the server 207 may forward the task to a region server 131 even if the POP 120 has sufficient computing resource capacity to execute the task. For example, the server 207 may use information stored in the user and task history data store 126 as well as information identifying how busy the POP 120 is, a time it may take to execute the task locally at the POP 120, a time it may take to execute the task within a regional data center 130, the time of day that the request is received, and/or the like to determine whether to execute the task locally or whether to make a determination as to which region server 131 should receive a forwarded task.

If the server 207 instructs the on-demand code execution environment 124 to execute the task, then the on-demand code execution environment 124 executes the task at (5). After execution, the on-demand code execution environment 124 reports results of the execution to the server 207 at (6).

If the server 207 does not decide to instruct the on-demand code execution environment 124 to execute the task, then the server 207 may query the task history from the user and task history data store 126 at (7). As described herein, the task history may include information identifying which regional data centers 130 have previously received requests to execute the task and/or have provisioned virtual machine instances to execute the task. In an embodiment, the server 207 further queries the user and task history data store 126 to identify which regional data centers 130 have stored a current version of the user-defined code used to execute the task. Using some or all of this information, the server 207 selects a geographic region to execute the task at (8).

Once a geographic region is selected, the server 207 forwards the task to the region server 131 in the selected geographic region at (9). The region server 131 then instructs the on-demand code execution environment 132 to execute the task at (10). In response to the instruction, the on-demand code execution environment 132 executes the task at (10) and reports results of the execution to the region server 131 at (12). The region server 131 then reports the execution results to the server 207 at (13).

Once the server 207 receives the execution results (either from the on-demand code execution environment 124 or from the region server 131), then the server 207 reports the execution results to the user device 102 at (14). Thus, if the POP 120 is unable to execute a task, the server 207 can use gathered metrics to identify a geographic region that can more efficiently execute the task.

Example Block Diagram for Replicating User-Defined Code to Other Regions

FIG. 4 is a block diagram of the operating environment 100 of FIG. 1 illustrating the operations performed by the components of the operating environment 100 to replicate user-defined code to one or more geographic regions, according to one embodiment. As illustrated in FIG. 4, the on-demand code execution environment 132A sends a notification to the replication system 133A indicating that user-defined code is created or updated at (1). For example, the user-defined code may be stored in the data store 260A in the on-demand code execution environment 132A and may be associated with the geographic region in which the regional data center 130A is present. The on-demand code execution environment 132A may monitor for changes to user-defined code or for the creation of new user-defined code at the on-demand code execution environment 132A (e.g., the frontend 220A) may notify the replication system 133A when this occurs so that the replication system 133A can replicate the code to other geographic regions.

Upon receiving the notification, the replication system 133A can retrieve the created or updated user-defined code from the on-demand code execution environment 132A at (2). Once the user-defined code is retrieved, the replication system 133A can replicate the user-defined code at (3). For example, the replication system 133A may replicate the bits of the user-defined code.

After the user-defined code is replicated, the replication system 133A may begin transmitting the replicated user-defined code to one or more other replication systems 133B-N at (4). For example, the replication system 133A may transmit the replicated user-defined code to the other replication systems 133B-N in parallel, in sequence, in bursts or groups, and/or in any combination thereof.

In other embodiments, not shown, the replication system 133A can alternatively or in addition transmit the replicated user-defined code to one or more POPs 120A-N. The POPs 120A-N may receive the replicated user-defined code such that these entities do not have to retrieve the user-defined code from the geographic region in which the code is stored and/or so that the POPs 120A-N can forward the replicated user-defined code to the various regional data centers 130B-N.

Periodically, the replication system 133A may determine a status of the transmissions of the replicated user-defined code. For example, the replication system 133A can transmit a message to any number of the replication systems 133B-N requesting the replication status at (5). The replication status can include a version of the user-defined code currently stored in the regional data center 130B-N, whether the latest replication transmission is complete (and if not, the progress of the transmission), and/or whether the replication transmission has started. The replication system 133A may store the received statuses in the replication data store 135A at (6).

Synchronously with or asynchronously from the replication status queries, the replication system 133A can retrieve the statuses stored in the replication data store 135A at (7) and transmit the statuses to the server 207 at (8). The replication system 133A may forward the statuses to the server 207A to aid the server 207A in making a determination of which geographic region to select to receive a forwarded task.

Example Table in the Replication Data Store

FIG. 5 is an example table 510 stored in the replication data store 135 of FIG. 1, according to one embodiment. The table 510 may be associated with a specific version of user-defined code that is associated with a particular geographic region. Thus, if multiple versions of the same user-defined code exist in a particular geographic region, then the replication data store 135 may include a separate table 510 for each version of the user-defined code.

As illustrated in FIG. 5, the table 510 identifies a name of the user-defined code (e.g., function 515), a current version of the user-defined code (e.g., version 1.6), geographic regions that are authorized to execute the task corresponding to the user-defined code (e.g., geographic regions 520, 522, and 524), the replication status of the user-defined code stored in each of the authorized geographic regions, and a time that each geographic region was last fully synched with the geographic region of the function 515 (e.g., the last time that a particular geographic region received a complete replicated version of the function 515). For example, the status for the geographic region 520 indicates that the user-defined code is fully replicated with the last complete sync occurring at 12:45:32 on 3/23/2016, the status for the geographic region 522 indicates that the replication is in progress and is 50% complete with the last complete sync occurring at 15:02:44 on 3/22/2016, and the status for the geographic region 524 indicates that the replication has not started and the version of the user-defined code stored in the regional data center 130 in the geographic region 524 is version 1.5 with the last complete sync occurring at 22:10:31 on 3/19/2016. The replication data store 135 may include additional tables, one for each user-defined code that is associated with the geographic region in which the replication data store 135 is present and/or one for each version of the user-defined code that is associated with the geographic region in which the replication data store 135 is present.

Example Task Execution Routines

FIG. 6 is a flow diagram depicting a task execution routine 600 illustratively implemented by a POP, according to one embodiment. As an example, the POP 120 (e.g., the server 207) of FIGS. 1 and 2A-2B can be configured to execute the task execution routine 600. The task execution routine 600 begins at block 602.

At block 604, a task is received. For example, the task may be received from a user device 102. The task may be received for the purpose of executing the task.

At block 606, a determination is made as to whether the POP 120 will execute the task locally. For example, the POP 120 determines whether there is a sufficient amount of computing resources available to execute the task locally (and may determine to execute the task locally if the POP 120 has sufficient computing resources available). As another example, the POP 120 determines whether requests to execute the task are commonly received (and may determine to execute the task locally if requests to execute the task are common). If the POP 120 determines to execute the task locally, the task execution routine 600 proceeds to block 608. Otherwise, if the POP 120 determines not to execute the task locally, the task execution routine 600 proceeds to block 612.

At block 608, the task is transmitted to the local code execution environment. For example, the task is transmitted to the on-demand code execution environment 124. The local code execution environment may then use a virtual machine instance to execute the task.

At block 610, an execution result is received from the local code execution environment. The execution result may include content (e.g., media, audio content, video content, etc.), a confirmation that an action corresponding to the task has been performed (e.g., a change to a parameter, the authentication of credentials, etc.), a notification that execution failed, and/or the like.

At block 612, the task is transmitted to a server in a geographic region. For example, the POP 120 transmits the task to the region server 131 in the geographic region closest to the POP 120. The server in the geographic region then forwards the task to a code execution environment local to the geographic region for execution. The results of the execution are then forwarded from the code execution environment local to the geographic region to the server in the geographic region.

At block 614, an execution result is received from the server in the geographic region. Thus, the POP 120 may receive the execution result from the local code execution environment or from a server in a geographic region.

At block 616, the execution result is transmitted. For example, the execution result is transmitted to the user device 102 that initiated the task call. After the execution result is transmitted, the task execution routine 600 may be complete, as shown in block 618.

FIG. 7 is another flow diagram depicting a task execution routine 700 illustratively implemented by a POP, according to one embodiment. As an example, the POP 120 (e.g., the server 207) of FIGS. 1 and 2A-2B can be configured to execute the task execution routine 700. The task execution routine 700 begins at block 702.

At block 704, a task is received. For example, the task may be received from a user device 102. The task may be received for the purpose of executing the task.

At block 706, a determination is made as to whether the POP 120 will execute the task locally. For example, the POP 120 determines whether there is a sufficient amount of computing resources available to execute the task locally (and may determine to execute the task locally if the POP 120 has sufficient computing resources available). As another example, the POP 120 determines whether requests to execute the task are commonly received (and may determine to execute the task locally if requests to execute the task are common). If the POP 120 determines to execute the task locally, the task execution routine 700 proceeds to block 708. Otherwise, if the POP 120 determines not to execute the task locally, the task execution routine 700 proceeds to block 712.

At block 708, the task is transmitted to the local code execution environment. For example, the task is transmitted to the on-demand code execution environment 124. The local code execution environment may then use a virtual machine instance to execute the task.

At block 710, an execution result is received from the local code execution environment. The execution result may include content (e.g., media, audio content, video content, etc.), a confirmation that an action corresponding to the task has been performed (e.g., a change to a parameter, the authentication of credentials, etc.), a notification that execution failed, and/or the like.

At block 712, a geographic region to execute the task is determined. While the default geographic region to execute the task may be the geographic region that is closest to the POP 120, the closest geographic region may not always be the most efficient geographic region to handle the task. For example, the POP 120 may consider which regional data centers 130 are storing a current version of the user-defined code, which regional data centers 130 have previously received a request to execute the task, and/or which regional data centers 130 have provisioned a virtual machine instance to execute the task. The POP 120 can, for example, determine which regional data centers 130 have a current version of the user-defined code stored therein and then select one regional data center 130 from those that have the current version of the user-defined code to execute the task. The POP 120 may consider which of those regional data centers 130 has previously received a request to execute the task or has a virtual machine instance provisioned to execute the task in selecting the appropriate regional data center 130. In further embodiments, the closest or otherwise most-efficient regional data center 130 is not selected because that regional data center 130 does not have a virtual machine instance provisioned to execute the task or otherwise has never previously received a request to execute the task. In such a situation, the POP 120 may instruct the regional data center 130 to provision a virtual machine instance to execute the task so that this regional data center 130 can receive future task execution requests.

At block 714, the task is transmitted to a server in the determined geographic region. The server in the determined geographic region then forwards the task to a code execution environment local to the determined geographic region for execution. The results of the execution are then forwarded from the code execution environment local to the determined geographic region to the server in the determined geographic region.

At block 716, an execution result is received from the server in the determined geographic region. Thus, the POP 120 may receive the execution result from the local code execution environment or from a server in the determined geographic region.

At block 718, the execution result is transmitted. For example, the execution result is transmitted to the user device 102 that initiated the task call. After the execution result is transmitted, the task execution routine 700 may be complete, as shown in block 720.

Terminology

All of the methods and tasks described herein may be performed and fully automated by a computer system. The computer system may, in some cases, include multiple distinct computers or computing devices (e.g., physical servers, workstations, storage arrays, cloud computing resources, etc.) that communicate and interoperate over a network to perform the described functions. Each such computing device typically includes a processor (or multiple processors) that executes program instructions or modules stored in a memory or other non-transitory computer-readable storage medium or device (e.g., solid state storage devices, disk drives, etc.). The various functions disclosed herein may be embodied in such program instructions, or may be implemented in application-specific circuitry (e.g., ASICs or FPGAs) of the computer system. Where the computer system includes multiple computing devices, these devices may, but need not, be co-located. The results of the disclosed methods and tasks may be persistently stored by transforming physical storage devices, such as solid state memory chips or magnetic disks, into a different state. In some embodiments, the computer system may be a cloud-based computing system whose processing resources are shared by multiple distinct business entities or other users.

Depending on the embodiment, certain acts, events, or functions of any of the processes or algorithms described herein can be performed in a different sequence, can be added, merged, or left out altogether (e.g., not all described operations or events are necessary for the practice of the algorithm). Moreover, in certain embodiments, operations or events can be performed concurrently, e.g., through multi-threaded processing, interrupt processing, or multiple processors or processor cores or on other parallel architectures, rather than sequentially.

The various illustrative logical blocks, modules, routines, and algorithm steps described in connection with the embodiments disclosed herein can be implemented as electronic hardware (e.g., ASICs or FPGA devices), computer software that runs on computer hardware, or combinations of both. Moreover, the various illustrative logical blocks and modules described in connection with the embodiments disclosed herein can be implemented or performed by a machine, such as a processor device, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. A processor device can be a microprocessor, but in the alternative, the processor device can be a controller, microcontroller, or state machine, combinations of the same, or the like. A processor device can include electrical circuitry configured to process computer-executable instructions. In another embodiment, a processor device includes an FPGA or other programmable device that performs logic operations without processing computer-executable instructions. A processor device can also be implemented as a combination of computing devices, e.g., a combination of a DSP and a microprocessor, a plurality of microprocessors, one or more microprocessors in conjunction with a DSP core, or any other such configuration. Although described herein primarily with respect to digital technology, a processor device may also include primarily analog components. For example, some or all of the rendering techniques described herein may be implemented in analog circuitry or mixed analog and digital circuitry. A computing environment can include any type of computer system, including, but not limited to, a computer system based on a microprocessor, a mainframe computer, a digital signal processor, a portable computing device, a device controller, or a computational engine within an appliance, to name a few.

The elements of a method, process, routine, or algorithm described in connection with the embodiments disclosed herein can be embodied directly in hardware, in a software module executed by a processor device, or in a combination of the two. A software module can reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disk, a removable disk, a CD-ROM, or any other form of a non-transitory computer-readable storage medium. An exemplary storage medium can be coupled to the processor device such that the processor device can read information from, and write information to, the storage medium. In the alternative, the storage medium can be integral to the processor device. The processor device and the storage medium can reside in an ASIC. The ASIC can reside in a user terminal. In the alternative, the processor device and the storage medium can reside as discrete components in a user terminal.

Conditional language used herein, such as, among others, “can,” “could,” “might,” “may,” “e.g.,” and the like, unless specifically stated otherwise, or otherwise understood within the context as used, is generally intended to convey that certain embodiments include, while other embodiments do not include, certain features, elements or steps. Thus, such conditional language is not generally intended to imply that features, elements or steps are in any way required for one or more embodiments or that one or more embodiments necessarily include logic for deciding, with or without other input or prompting, whether these features, elements or steps are included or are to be performed in any particular embodiment. The terms “comprising,” “including,” “having,” and the like are synonymous and are used inclusively, in an open-ended fashion, and do not exclude additional elements, features, acts, operations, and so forth. Also, the term “or” is used in its inclusive sense (and not in its exclusive sense) so that when used, for example, to connect a list of elements, the term “or” means one, some, or all of the elements in the list.

Disjunctive language such as the phrase “at least one of X, Y, or Z,” unless specifically stated otherwise, is otherwise understood with the context as used in general to present that an item, term, etc., may be either X, Y, or Z, or any combination thereof (e.g., X, Y, or Z). Thus, such disjunctive language is not generally intended to, and should not, imply that certain embodiments require at least one of X, at least one of Y, and at least one of Z to each be present.

While the above detailed description has shown, described, and pointed out novel features as applied to various embodiments, it can be understood that various omissions, substitutions, and changes in the form and details of the devices or algorithms illustrated can be made without departing from the spirit of the disclosure. As can be recognized, certain embodiments described herein can be embodied within a form that does not provide all of the features and benefits set forth herein, as some features can be used or practiced separately from others. The scope of certain embodiments disclosed herein is indicated by the appended claims rather than by the foregoing description. All changes which come within the meaning and range of equivalency of the claims are to be embraced within their scope.

Claims

1. A computer-implemented method of selecting a location at which to execute user-defined code, the method comprising: receiving a first set of user-defined code that is generated by a user for execution on a first on-demand code execution environment at a point of presence (POP), wherein the first on-demand code execution environment comprises a plurality of virtual machine instances executing an initialized operating system and awaiting requests to execute user-defined code within that operating system;replicating the first set of user-defined code to a first set of geographic regions in a plurality of geographic regions;receiving, from a user device associated with the user that generated the first set of user-defined code, a request to execute a task, wherein execution of the task comprises execution of the first set of user-defined code, and wherein the request comprises an argument for use in execution of the first set of user-defined code;determining that the task should not be executed by the first on-demand code execution environment;determining that a current version of the first set of user-defined code is replicated to the first set of geographic regions in the plurality of geographic regions;selecting a first geographic region in the first set of geographic regions to execute the task based at least in part on requests to execute the task previously received by a second on-demand code execution environment located in the first geographic region;transmitting the request to execute the task to the first geographic region such that the task is executed by the second on-demand code execution environment, wherein the first on-demand code execution environment is located at a geographic location different than the first geographic region; andtransmitting results of execution of the task to the user device.

2. The computer-implemented method of claim 1, wherein the first set of geographic regions comprises a second geographic region, and wherein the method further comprises instructing the second geographic region to provision a virtual machine instance for executing the task.

3. The computer-implemented method of claim 2, further comprising: receiving an indication that the virtual machine instance is provisioned;receiving, from a second user device, a second request to execute the task;determining that the task should not be executed by the first on-demand code execution environment;selecting the second geographic region to execute the task;transmitting the second request to execute the task to the second geographic region such that the task is executed by a third on-demand code execution environment; andtransmitting second results of execution of the task to the second user device.

4. The computer-implemented method of claim 1, wherein the second on-demand code execution environment comprises a first virtual machine instance that executes the task by processing the current version of the first set of user-defined code.

5. The computer-implemented method of claim 1, wherein that the task should not be executed by the first on-demand code execution environment further comprises determining that the task should not be executed by the first on-demand code execution environment based on at least one of an amount of computing resources available to the one or more computing devices, how often the one or more computing devices are requested to execute the task, a duration of time to execute the task by the first on-demand code execution environment, a duration of time to execute the task by the second on-demand code execution environment, a historical volume of requests received from the user device, a time of day that the request to execute the task is received, a latency-sensitivity of the task, or properties of the task.

6. A system for selecting a location at which to execute user-defined code, the system comprising: a first on-demand code execution environment that comprises a first computing device configured to execute code at a point of presence (POP); anda second computing device comprising a processor in communication with the first on-demand code execution environment and configured with specific computer-executable instructions to: receive a first set of user-defined code that is generated by a user for execution on the first on-demand code execution environment;receive, from a user device associated with the user that generated the first set of user-defined code, a request to execute a task, wherein execution of the task comprises execution of the first set of user-defined code;determine that the task should not be executed by the first on-demand code execution environment;receive an indication that a copy of the first set of user-defined code is replicated to a first set of geographic regions in a plurality of geographic regions;select a first geographic region in the first set of geographic regions to execute the task based at least in part on the first geographic region having a current version of the first set of user-defined code, wherein a second on-demand code execution environment is located in the first geographic region;transmit the request to execute the task to the first geographic region such that the task is executed by the second on-demand code execution environment; andtransmit results of execution of the task to the user device.

7. The system of claim 6, wherein the first set of geographic regions comprises a second geographic region, and wherein second computing device is further configured with specific computer-executable instructions to instruct the second geographic region to provision a virtual machine instance for executing the task.

8. The system of claim 7, wherein second computing device is further configured with specific computer-executable instructions to: receive an indication that the virtual machine instance is provisioned;receive, from a second user device, a second request to execute the task;determine that the task should not be executed by the first on-demand code execution environment;select the second geographic region to execute the task;transmit the second request to execute the task to the second geographic region such that the task is executed by a third on-demand code execution environment; andtransmit second results of execution of the task to the second user device.

9. The system of claim 6, wherein the first computing device is at least one of a virtual machine instance or a physical computing device.

10. The system of claim 6, wherein the second on-demand code execution environment comprises a first virtual machine instance that executes the task by processing the copy of the first set of user-defined code.

11. The system of claim 6, wherein the second computing device is further configured with specific computer-executable instructions to determine that the task should not be executed by the first on-demand code execution environment based on at least one of an amount of computing resources available to the system, how often the system is requested to execute the task, a duration of time to execute the task by the first on-demand code execution environment, a duration of time to execute the task by the second on-demand code execution environment, a historical volume of requests received from the user device, a time of day that the request to execute the task is received, a latency-sensitivity of the task, or properties of the task.

12. The system of claim 6, wherein the second computing device is further configured with specific computer-executable instructions to select the first geographic region in the first set of geographic regions to execute the task based on at least one of the first geographic region having a current version of the user-defined code stored therein, the first geographic region having a virtual machine instance provisioned to execute the task, or the first geographic region having previously received a second request to execute the task.

13. The system of claim 6, wherein the POP services the first geographic region.

14. The system of claim 13, wherein the POP services the first geographic region and a second geographic region in the plurality of geographic regions, and wherein a third on-demand code execution environment is located in the second geographic region.

15. The system of claim 14, wherein the first set of user-defined code is associated with a third geographic region in the plurality of geographic regions, and wherein the current version of the first set of user-defined code has not been replicated to the second geographic region.

16. Non-transitory, computer-readable storage media comprising computer-executable instructions for selecting a location at which to execute code, wherein the computer-executable instructions, when executed by a computer system at a point of presence (POP), cause the computer system to: receive a first set of user-defined code that is generated by a user for execution on a first on-demand code execution environment, wherein the first on-demand code execution environment comprises a first computing device configured to execute code;receive a request to execute a task, wherein execution of the task comprises execution of the first set of user-defined code;determine that the task should not be executed by the first on-demand code execution environment;receive an indication that a copy of the first set of user-defined code is replicated to a first set of geographic regions in a plurality of geographic regions;select a first geographic region in the first set of geographic regions to execute the task based at least in part on the first geographic region having a current version of the first set of user-defined code, wherein a second on-demand code execution environment is located in the first geographic region; andtransmit the request to execute the task to the first geographic region such that the task is executed by the second on-demand code execution environment.

17. The non-transitory, computer-readable storage media of claim 16, wherein the first set of geographic regions comprises a second geographic region, and wherein the computer-executable instructions further cause the computer system to: instruct the second geographic region to provision a virtual machine instance for executing the task;receive an indication that the virtual machine instance is provisioned;receive, from a second user device, a second request to execute the task;determine that the task should not be executed by the first on-demand code execution environment;select the second geographic region to execute the task;transmit the second request to execute the task to the second geographic region such that the task is executed by a third on-demand code execution environment; andtransmit second results of execution of the task to the second user device.

18. The non-transitory, computer-readable storage media of claim 16, wherein the computer-executable instructions further cause the computer system to transmit results of execution of the task to a user device.

19. The non-transitory, computer-readable storage media of claim 16, wherein the POP services the first geographic region and a second geographic region in the plurality of geographic regions.

20. The non-transitory, computer-readable storage media of claim 19, wherein the first set of user-defined code is associated with a third geographic region in the plurality of geographic regions, and wherein the current version of the first set of user-defined code has not been replicated to the second geographic region.