This application generally relates to data storage systems, and more particularly to use of SCM on a data storage system.
Data storage systems (often referred to herein simply as “storage systems”) may include storage resources used by one or more host systems (sometimes referred to herein as “hosts”), i.e., servers, to store data. One or more storage systems and one or more host systems may be interconnected by one or more network components, for example, as part of a switching fabric, to form a data storage network (often referred to herein simply as “storage network”). Storage systems may provide any of a variety of data services to host systems of the storage network.
A host system may have host applications that utilize the data services provided by one or more storage systems of the storage network to store data on the physical storage devices (e.g., tape, disks or solid state devices) thereof. For a given application, to perform input/output (IO) operations utilizing a physical storage device of the storage system, one or more components of the host system, storage system and network components therebetween may be used.
Host systems may not address the physical storage devices (e.g., disk drives or flash drives) of a storage systems directly, but rather access to data may be provided to one or more host systems from what the host systems view as a plurality of logical storage units (LSUs) including, for example, logical blocks, logical devices (also referred to as logical volumes, LUNs, logical storage units and/or logical disks), thin devices, groups of logical devices (e.g., storage groups), NVMe namespaces, and other types of LSUs. LSUs are described in more detail elsewhere herein.
In some embodiments of the invention, non-volatile random access memory on a data storage is used, including, using a first portion of the non-volatile random access memory as temporary memory, converting at least a first sub-portion of the first portion of non-volatile random access memory to use as persistent storage, and, in response to an I/O request, accessing the first sub-portion as persistent storage. The I/O request may be a write request that specifies data associated with a virtual memory location, and accessing the sub-portion may include, in response to the write request, allocating the first sub-portion to the virtual memory location and writing the data to the sub-portion. The allocation may be performed by a CPU, and the data storage system may include a data structure external to the CPU that maps virtual memory locations to locations in the non-volatile random access memory, the method further may include creating an entry in the data structure that maps the virtual memory location to the first sub-portion. The method may further include receiving another I/O request specifies data associated with a virtual memory location, and accessing the data structure independently of the CPU to determine that the virtual memory location is mapped to the first sub-portion. The method may further include queuing one or more sub-portions for allocation in response to write requests. The method may further include allocating the one or more sub-portions in response to write requests, determining that at least one allocated sub-portion no longer stores data, and reclaiming the at least one allocated sub-portion for further allocation. The data storage system may include an internal switching fabric to which the non-volatile random access memory is connected, and a host system may include a storage system interface directly connected to the switching fabric independent of any director, and the I/O request may be sent directly from the storage system interface over the internal fabric independent of a director and/or an I/O protocol. The non-volatile random access memory may be embodied as one or more dual in-line memory modules.
In other embodiments of the invention, a data storage system includes non-volatile random access memory, and executable logic that implements a method including using a first portion of the non-volatile random access memory as temporary memory, converting at least a first sub-portion of the first portion of non-volatile random access memory to use as persistent storage, and, in response to an I/O request, accessing the first sub-portion as persistent storage. The I/O request may be a write request that specifies data associated with a virtual memory location, and wherein accessing the sub-portion may include, in response to the write request, allocating the first sub-portion to the virtual memory location, and writing the data to the sub-portion. The allocation may be performed by a CPU, and the data storage system may include a data structure external to the CPU that maps virtual memory locations to locations in the non-volatile random access memory, and the method further may include creating an entry in the data structure that maps the virtual memory location to the first sub-portion. The method further may include receiving another I/O request specifies data associated with a virtual memory location; and accessing the data structure independently of the CPU to determine that the virtual memory location is mapped to the first sub-portion. The method further may include queuing one or more sub-portions for allocation in response to write requests. The method further may include allocating the one or more sub-portions in response to write requests, determining that at least one allocated sub-portion no longer stores data and reclaiming the at least one allocated sub-portion for further allocation. The data storage system may include an internal switching fabric to which the non-volatile random access memory is connected, and a host system may include a storage system interface directly connected to the switching fabric independent of any director; and the I/O request may be sent directly from the storage system interface over the internal fabric independent of a director and/or an I/O protocol. The non-volatile random access memory may be embodied as one or more dual in-line memory modules.
In another embodiment, computer-readable media for a data storage system including non-volatile random access memory may be provided. The computer-readable media has software stored thereon including executable code that uses a first portion of the non-volatile random access memory as temporary memory, executable code that converts at least a first sub-portion of the first portion of non-volatile random access memory to use as persistent storage and executable code that accesses, in response to an I/O request, the first sub-portion as persistent storage. The software further may include executable code that allocates the one or more sub-portions in response to write requests, executable code that determines that at least one allocated sub-portion no longer stores data and executable code that reclaims the at least one allocated sub-portion for further allocation. The data storage system may include an internal switching fabric to which the non-volatile random access memory is connected, and a host system may include a storage system interface directly connected to the switching fabric independent of any director, and the I/O request may be sent directly from the storage system interface over the internal fabric independent of a director and/or an I/O protocol. The non-volatile random access memory may be embodied as one or more dual in-line memory modules.
Features and advantages of the present invention will become more apparent from the following detailed description of illustrative embodiments thereof taken in conjunction with the accompanying drawings in which:
A host system (i.e., server) may include one or more CPUs that each has its own internal volatile memory used for internal functions of the CPU (e.g., registers and cache), typically implemented with SRAM memory technology. The typical main memory of a host system (external to the CPU(s)) used for shorter-term, temporary storing of data while executing applications, often referred to as “RAM,” is typically implemented using volatile DRAM memory technology, which is slower and less expensive than SRAM, but faster and more expensive than non-volatile random-access memory (NVRAM). NVRAM is often referred to herein as “NVR.” The capacity of the main memory may be increased by adding one or more physically discrete memory components (e.g., memory cards) to the host system. The added memory may be any of a variety of types of memory, including volatile memory (e.g., DRAM) or NVR. NVR may include flash memory as well as more recently developed storage class memory (SCM) including, for example, 3D XPoint (e.g., Intel® Optane™ memory) and resistive RAM (ReRAM)). NVR is slower and typically cheaper than DRAM, albeit SCM technology has significantly reduced the differences in speed.
When the added memory is of a different type than the native main memory (e.g., DRAM) of the host system, for example, when the added memory is NVR, the operating system (OS) of the host and/or one or more applications executing on the host system, may be configured to be aware of the different types of the memory and to utilize the native main memory (often referred to herein as “native memory”) and added memory differently based on their respective properties, including speed (i.e., access time) and volatility (i.e., being volatile or non-volatile). For example, the added memory may be treated as a second tier of memory—the native memory (and perhaps other added memory of the same type as the native memory) being the first tier. A second tier of memory may be referred to herein as “2T memory.” In the case where the 2T memory is NVR, the 2T memory may be utilized as persistent memory for which data stored thereon (including state information) persists (i.e., remains available) after the process or program that created the data terminates, perhaps even after the host system and/or the 2T memory components fails. A common application for which 2T NVR may be employed as persistent memory is an in-memory database, e.g., made available by Oracle Corporation and other companies.
A potential drawback of added memory (i.e., memory added to the native memory of a host system or other computer)—whether treated as 2T memory or otherwise—is that it is static; i.e., fixed in size such that the capacity is limited by the amount of memory included on the physical component (e.g., memory card). As a result, if the amount needed by a host system—e.g., to support the one or more applications executing on it—exceeds the memory capacity of the physical component, then an additional physical memory component may need to be acquired, installed, configured and allocated. On the other hand, the memory capacity of the added physical memory component may exceed (perhaps significantly) the memory demands of the host system, in which case much of the capacity (and the money spend to acquire it) may be wasted.
Because of the foregoing drawbacks and other drawbacks, it may be desirable to expand the memory capacity of a host system by making available the memory of a storage system for use as host memory of one or more hosts, for example, as an extension of the main memory of one or more hosts, e.g., as described in U.S. patent application Ser. No. 16/864,824, titled “Using Storage System as Host Memory,” to Jon Krasner et al., the entire contents of which is hereby incorporated by reference in its entirety.
In some embodiments, it may be desirable to use NVR (e.g., SCM) of the storage system memory, for example, e.g., embodied as a dual in-line memory module (DIMM), as persistent memory of the host system, e.g., in a storage context. However, it also may be desirable for the storage system to use its NVR as byte-addressable memory, for example, in a memory context for which volatile memory (e.g., DRAM) is typically used. For example, it may be desirable for the storage system to use NVR for temporary storage, e.g., for caching data being written to and/or read from the storage system by hosts. Accordingly, it may be desirable to enable NVR of a storage system to be used as both as memory and as storage.
Described herein are techniques and mechanisms for dynamically converting NVR on a storage system between use as temporary memory in a memory context and persistent memory in a storage context. NVR being used as temporary memory in a memory context may be referred to herein as memory or as being used as memory. Similarly, NVR being used as persistent memory in a storage context may be referred to herein as storage or as being used as storage. In some embodiments of the invention, NVR (e.g., embodied as DIMM) may be utilized in a hybrid capacity, where some of the NVR is used as memory (“NVR memory”) and some of the NVR is used as storage (“NVR storage”), and where NVR memory is converted to memory as needed, dynamically as I/O is being processed using the NVR.
In some embodiments of the invention, a host system may be directly connected to an internal switching fabric (“internal fabric”) of the data storage system without an intervening component of the storage system (e.g., a director) controlling access of the host system to the internal fabric or to the memory and physical devices connected to the internal fabric. For example, the host system may include a storage system interface (SSI) physically connected to a remainder of the host system by one or more interconnects or busses (e.g., in accordance with PCIe technology), where the SSI may be physically connected directly to the internal fabric of the data storage system. For example, the SSI may reside on an integrated circuit card or other physical component that is plugged into a port of the host system. The direct connection to the internal switching fabric may enable the host system to bypass networks external to the storage system (including switch fabrics and physical links), and bypass internal components of the storage system, including directors and the like, when accessing NVR (e.g., embodied as DIMM) or physical storage devices of the storage system. Such bypassing may enable hosts to communicate with storage system memory and physical storage devices without using one or more proprietary or standard protocols (e.g., Fibre Channel (FC), SCSI, iSCSI) required by the bypassed external networks and/or internal components of the storage system, and avoid the processing delays inherent in the bypassed switch fabrics, directors and other components processing communications in accordance with such protocols. Thus, the host system being directly connected to the internal fabric of the storage system may reduce (e.g., substantially) I/O (e.g., read and write) response times and access times to other information (e.g., metadata) on the storage system.
In embodiments in which a host system is directly connected to an internal fabric of the data storage system, the host system may provision and use the NVR (e.g., embodied as DIMM) as storage by directly communicating with the NVR over the internal fabric, for example, using RDMA, as described in more detail elsewhere herein. The logic for provisioning the NVR as memory or storage, converting NVR memory to NVR storage and allocating NVR storage may reside on the storage system and/or on one or more host systems, for example, on an SSI. When such logic resides on the storage system, the host system (e.g., through an SSI) may access such logic directly over the internal fabric, for example, using RDMA.
It should be appreciated that while embodiments of the invention are described as using SCM as NVR, the invention is not so limited, as other types of NVR may be employed.
Illustrative embodiments of the invention will now be described in more detail in relation to the figures.
The N hosts 14a-n may access the storage system 20a, for example, in performing input/output (IO) operations or data requests, through network 18. For example, each of hosts 14a-n may include one or more host bus adapters (HBAs) (not shown) that each include one or more host ports for connecting to network 18. The network 18 may include any one or more of a variety of communication media, switches and other components known to those skilled in the art, including, for example: a repeater, a multiplexer or even a satellite. Each communication medium may be any of a variety of communication media including, but not limited to: a bus, an optical fiber, a wire and/or other type of data link, known in the art. The network 18 may include at least a portion of the Internet, or a proprietary intranet, and components of the network 18 or components connected thereto may be configured to communicate in accordance with any of a plurality of technologies, including, for example: SCSI, ESCON, Fibre Channel (FC), iSCSI, FCoE, GIGE (Gigabit Ethernet), NVMe over Fabric (NVMeoF); other technologies, or any suitable combinations of the foregoing, each of which may have one or more associated standard specifications. In some embodiments, the network 18 may be, or include, a storage network fabric including one or more switches and other components. A network located externally to a storage system that connects host systems to storage system resources of the storage system, may be referred to herein as an “external network.”
Each of the host systems 14a-n and the storage systems 20a-n included in the storage network 10 may be connected to the network 18 by any one of a variety of connections as may be provided and supported in accordance with the type of network 18. The processors included in the host computer systems 14a-n may be any one of a variety of proprietary or commercially available single or multi-processor system, such as an Intel-based processor, or other type of commercially available processor able to support traffic in accordance with each particular embodiment and application. Each of the host computer systems may perform different types of I/O operations in accordance with different tasks and applications executing on the hosts. In the embodiment of
Each of the storage systems 20a-n may be manufactured by different vendors and interconnected (not shown). Additionally, the storage systems 20a-n also may be connected to the host systems through any one or more communication connections 31 that may vary with each particular embodiment and device in accordance with the different protocols used in a particular embodiment. The type of communication connection used may vary with certain system parameters and requirements, such as those related to bandwidth and throughput required in accordance with a rate of I/O requests as may be issued by each of the host computer systems 14a-n, for example, to the storage systems 20a-20n. It should be appreciated that the particulars of the hardware and software included in each of the components that may be included in the storage systems 20a-n are described herein in more detail, and may vary with each particular embodiment.
Each of the storage systems, such as 20a, may include a plurality of physical storage devices 24 (e.g., physical non-volatile storage devices) such as, for example, disk devices, solid-state storage devices (SSDs, e.g., flash, storage class memory (SCM), NVMe SSD, NVMe SCM) or even magnetic tape, and may be enclosed within a disk array enclosure (DAE) 27. In some embodiments, two or more of the physical storage devices 24 may be grouped or arranged together, for example, in an arrangement consisting of N rows of physical storage devices 24a-n. In some embodiments, one or more physical storage devices (e.g., one of the rows 24a-n of physical storage devices) may be connected to a back-end adapter (“BE”) (e.g., a director configured to serve as a BE) responsible for the backend management of operations to and from a portion of the physical storage devices 24. A BE is sometimes referred to by those in the art as a disk adapter (“DA”) because of the development of such adapters during a period in which disks were the dominant type of physical storage device used in storage systems, even though such so-called DAs may be configured to manage other types of physical storage devices (e.g., SSDs). In the system 20a, a single BE, such as 23a, may be responsible for the management of one or more (e.g., a row) of physical storage devices, such as row 24a. That is, in some configurations, all I/O communications with one or more physical storage devices 24 may be controlled by a specific BE. BEs 23a-n may employ one or more technologies in communicating with, and transferring data to/from, physical storage devices 24, for example, SAS, SATA or NVMe. For NVMe, to enable communication between each BE and the physical storage devices that it controls, the storage system may include a PCIe switch for each physical storage device controlled by the BE; i.e., connecting the physical storage device to the controlling BE.
It should be appreciated that the physical storage devices are not limited to being arranged in rows. Further, the DAE 27 is not limited to enclosing disks, as the name may suggest, but may be constructed and arranged to enclose a plurality of any type of physical storage device, including any of those described herein, or combinations thereof.
The system 20a also may include one or more host adapters (“HAs”) 21a-n, which also are referred to herein as front-end adapters (“FAs”) (e.g., directors configured to serve as FAs). Each of these FAs may be used to manage communications and data operations between one or more host systems and GM 25b of memory 26. The FA may be, or include, a Fibre Channel (FC) adapter if FC is a technology being used to communicate between the storage system 20a and the one or more host systems 14a-n, or may be another type of adapter based on the one or more technologies being used for I/O communications.
Also shown in the storage system 20a is a remote adapter (“RA”) 40. The RA may be, or include, hardware that includes a processor used to facilitate communication between storage systems (e.g., 20a and 20n), such as between two of the same or different types of storage systems, and/or may be implemented using a director.
Storage system 20a also may include a management module 22, which may be configured (e.g., dedicated) to performing storage management functions or services such as, for example, storage provisioning, device configuration, tier management, other services, or any combination of other services. The management module may be configured to be accessed by only certain personnel (e.g., storage administrators, support engineers) and may have its own dedicated hardware, firmware, software, CPU resources and OS, and may be loaded with one or more applications, tools, CLIs, APIs and the like to enable management. In some embodiments, the management module, or portions thereof, may be located external to storage system 20a, for example, as part of one of host systems 14a-n or another separate system connected to storage system 20a via network 18.
The FAs, BEs and RA may be collectively referred to herein as directors 37a-n. Each director 37a-n may be implemented (e.g., in hardware, firmware, software or a combination thereof) on a circuit board that includes memory resources (e.g., at least a segment of GM portion 25b) and compute resources, for example, one or more processing cores (e.g., as part of a CPU) and/or a CPU complex for processing I/O operations, and that as described in more detail elsewhere herein. There may be any number of directors 37a-n, which may be limited based on any of a number of factors, including spatial, computation and storage limitations. In an embodiment disclosed herein, there may be up to sixteen directors coupled to the memory 26. Other embodiments may use a higher or lower maximum number of directors.
System 20a also may include an internal switching fabric (i.e., internal fabric) 30, which may include one or more switches, that enables internal communications between components of the storage system 20a, for example, directors 37a-n (FAs 21a-n, BEs 23a-n, RA 40, management module 22) and memory 26, e.g., to perform I/O operations. One or more internal logical communication paths may exist between the directors and the memory 26, for example, over the internal fabric 30. For example, any of the directors 37a-n may use the internal fabric 30 to communicate with other directors to access any of physical storage devices 24; i.e., without having to use memory 26. In addition, one of the directors 37a-n may be able to broadcast a message to all of the other directors 37a-n over the internal fabric 30 at the same time. Each of the components of system 20a may be configured to communicate over internal fabric 30 in accordance with one or more technologies such as, for example, InfiniBand (TB), Ethernet, Gen-Z, another technology, or any suitable combination of the foregoing.
The GM portion 25b may be used to facilitate data transfers and other communications between the directors 37a-n in a storage system. In one embodiment, the directors 37a-n (e.g., serving as FAs or BEs) may perform data operations using a cache 28 that may be included in the GM 25b, for example, in communications with other directors, and other components of the system 20a. The other portion 25a is that portion of memory that may be used in connection with other designations that may vary in accordance with each embodiment. Global memory 25b and cache 28 are described in more detail elsewhere herein. It should be appreciated that, although memory 26 is illustrated in
In at least one embodiment, write data received at the storage system from a host or other client may be initially written to cache 28 and marked as write pending. For example, cache 28 may be partitioned into one or more portions called cache slots (which also may be referred to in the field of data storage as cache lines, cache blocks or another name), which may be a of a predefined uniform size, for example, 128 Kbytes. Write data of a write operation received at the storage system may be initially written (i.e., staged) in one or more of these cache slots and marked as write pending. Once written to cache 28, the host (e.g., one of 14a-n) may be notified that the write operation has completed. At a later time, the write data may be de-staged from cache 28 to one or more physical storage devices 24a-n, such as by a BE.
The memory 26 may include persistent memory for which for which data stored thereon persists after the process or program that created the data terminates. For example, at least portions of the memory 26 may be implemented using DIMM (or another type of fast RAM memory) that is battery-backed by a NAND-type memory (e.g., flash). In some embodiments, the data in such persistent memory may persist (for at least some period of time) after the storage system fails. The memory 26 (or at least a portion thereof—e.g., the cache 28 or a portion thereof) may be configured such that each data written to the memory 28 is mirrored to provide a form of write protection. For example, each memory location within each such mirrored portion of the memory 26 may have a corresponding memory location on the storage system 20a to which a redundant copy of the data is stored, and which can be used in place of the mirrored memory location in the event the mirrored memory location fails. The redundant memory location should be located outside of at least the most local fault zone of the mirrored memory location. In some embodiments described in more detail herein, the memory 26 may be distributed among multiple physically discrete processing nodes (e.g., circuit boards), in which case mirroring may be configured such that a mirrored memory location and its corresponding redundant memory location are located on different physically discrete processing nodes.
Storage system 20a may include a back-up power supply 41 (e.g., a battery) that can provide power to the storage system for a limited amount of time to after primary (AC) power fails. This limited time may allow certain tasks to be performed during a window of time beginning when the primary power fails until the earliest of: the primary power is restored; and the end of the limited lifetime (sometimes on the order of second or tens of seconds) of the back-up power supply. For example, the storage system 20a (e.g., the memory 26 and/or memory management module 32) may be configured to automatically copy the contents of the memory 26 during this window of time to one or more predetermined physical storage devices, to be restored to the memory 26 after the power has been restored, e.g., as part of the storage system recovering process. Such automatic copying for restoration during recovering may referred to herein as “vaulting.” Vaulting may provide a form of write protection for data written to the memory 26, for example, for dirty data in the cache 28; i.e., data written to the storage system, which has been staged in the cache 28 but not yet de-staged to a physical storage device. More broadly, vaulting may be performed for any data written to the memory 26.
The storage system 20a may include a memory management module 32 configured to manage one or more aspects of the memory 26, and the memory management module 32 may include a cache management module 34 for managing one or more aspects of the cache 28.
It should be noted that, although examples of techniques herein may be made with respect to a physical storage system and its physical components (e.g., physical hardware for each RA, BE, FA and the like), techniques herein may be performed in a physical storage system including one or more emulated or virtualized components (e.g., emulated or virtualized ports, emulated or virtualized BEs or FAs), and also a virtualized or emulated storage system including virtualized or emulated components. For example, in embodiments in which NVMe technology is used to communicate with, and transfer data between, a host system and one or more FAs, one or more of the FAs may be implemented using NVMe technology as an emulation of an FC adapter.
Any of storage systems 20a-n, or one or more components thereof, described in relation to
Host systems 14a-n may provide data and control (e.g., management and access control) information to storage systems 20a-n over a plurality of I/O paths defined between the host systems and storage systems, for example, including host system components, storage system components, and network components (e.g., of network 18), and the storage systems also may provide data to the host systems across the I/O paths. In the embodiment of
In the following paragraphs, further details are described with reference to processing node 212a but each of the N processing nodes in a system may be similarly configured. For example, processing node 212a may include any of: one or more directors 216a (e.g., directors 37a-n); memory portion 214a; one or more processing cores 217a including compute resources, for example, as part of a CPUs and/or a CPU complex for processing I/O operations; and a fabric interface module (FIM) 215a for interfacing the processing node 212a to an internal fabric 230. Each director 216a may be configured to operate, such as by executing code, as any one or more of an FA, BE, RA, and the like. In some embodiments, each of the directors, or a portion thereof, are implemented in software stored in a memory portion 214a (e.g., in a dedicated local memory 222a) that is executed by one or more of the processing cores 217a. Such software implementation of directors may be considered emulations of types of physical directors (i.e., directors implemented (at least primarily) in hardware).
Each FIM 215a-n may include one or more host channel adapters (HCAs) that physically couple, and are configured to enable communication between, its respective processing node 212a-n, and the internal fabric 230. In some embodiments, the internal fabric 230 may include multiple (e.g., 2) switches, and each HCA 215a-n may have multiple (e.g., 2) ports, each one connected directly to one of the switches.
Each of the processing nodes 212a-n may, respectively, also include memory portions 214a-n. The memory portion of each processing node may be characterized as locally accessible with respect to that particular processing node, and more specifically with respect to other components on the same processing node. For example, processing node 212a includes memory portion 214a which is memory that is local to that particular processing node 212a. Data stored in memory portion 214a may be directly accessed by any of the processing cores 217a (e.g., executing instructions on behalf of one of the directors 216a) of the processing node 212a. For example, memory portion 214a may be a fast memory (e.g., DIMM (dual inline memory module) DRAM (dynamic random access memory)) that is locally accessible by a director 216a, where data from one location in 214a may be copied to another location in 214a directly using DMA operations (e.g., local memory copy operations) issued by director 216a. Thus, the director 216a may directly access data of 214a locally without communicating over the fabric 230.
The memory portions 214a-214n of processing nodes 212a-n may be further partitioned into different portions or segments for different uses. For example, each of the memory portions 214a-214n may respectively include GM segments 220a-n configured for collective use as segments of a distributed GM, for example, GM 225 (e.g., GM 25b). Thus, data stored in any GM segment 220a-n may be accessed by any director 216a-n on any processing node 212a-n. Additionally, each of the memory portions 214a-n may respectively include dedicated local memories 222a-n. Each of the dedicated local memories 222a-n are respectively configured for use locally by the one or more directors 216a-n, and possibly other components, residing on the same single processing node. In at least one embodiment where there is a single director denoted by 216a (and generally by each of 216a-n), data stored in the dedicated local memory 222a may be accessed by the respective single director 216a located on the same processing node 212a. However, the remaining directors located on other ones of the N processing nodes may not access data stored in the dedicated local memory 222a.
To further illustrate, GM segment 220a may include information such as user data stored in the cache portion 220a, metadata, and the like, that is accessed (e.g., for read and/or write) generally by any director of any of the processing nodes 212a-n. Thus, for example, any director 216a-n of any of the processing nodes 212a-n may communicate over the fabric 230 to access data in GM segment 220a. In a similar manner, any director 216a-n of any of the processing nodes 212a-n may generally communicate over fabric 230 to access any GM segment 220a-n of the distributed GM. Although a particular GM segment, such as 220a, may be locally accessible to directors on one particular processing node, such as 212a, any director of any of the processing nodes 212a-n may generally access the GM segment 220a. Additionally, the director 216a also may use the fabric 230 for data transfers to and/or from GM segment 220a even though 220a is locally accessible to director 216a (without having to use the fabric 230).
Also, to further illustrate, dedicated local memory 222a may be a segment of the memory portion 214a on processing node 212a configured for local use solely by components on the single/same processing node 212a. For example, dedicated local memory 222a may include data described in following paragraphs which is used and accessed only by directors 216a included on the same processing node 212a as the dedicated local memory 222a. In at least one embodiment in accordance with techniques herein and as described elsewhere herein, each of the dedicated local memories 222a-n may include a local page table or page directory used, respectively, by only director(s) 216a-n local to each of the processing nodes 212a-n.
In such an embodiment as in
Each cache portion 254a-n may be a portion of a shared cache 228 (e.g., cache 28) distributed across the processing nodes 212a-n, where the shared cache 228 may be considered a part of the GM 225. The cache portion 254a-n may include a plurality of cache slots 256a-n, each cache slot including one or more (e.g., 16) sections 258a-n. Each cache slot 256a-n may be of a uniform size (e.g., 128 KB) and each section may be of a uniform size (e.g., 8 KB). It should be appreciated that cache slot sizes and section sizes other than 128 KB and 8 KB, and a quantity of sections other than 16, may be used.
In an embodiment, the storage system as described may be characterized as having one or more logical mapping layers in which an LSU of the storage system is exposed to the host whereby the LSU is mapped by such mapping layers of the storage system to one or more physical storage devices. Additionally, the host also may have one or more additional mapping layers so that, for example, a host-side LSU may be mapped to one or more storage system LSUs as presented to the host.
Any of a variety of data structures may be used to process I/O on storage system 20a, including data structures to manage the mapping of LSUs and locations thereon to physical storage devices and locations thereon. Such data structures may be stored in any of memory 26, including GM 25b and memory 25a, GM segment 220a-n and/or dedicated local memories 22a-n. Thus, storage system 20a, and storage system 620a described in more detail elsewhere herein, may include memory elements (e.g., cache) that hold data stored on physical storage devices or that is currently held (“staged”) and will be stored (“de-staged”) to physical storage devices, and memory elements that store metadata (e.g., any of the metadata described herein) associated with such data. Illustrative examples of data structures for holding such metadata will now be described.
Each of the entries 66-68 of the master LSU table 62 may correspond to, and include a reference to, another table corresponding to the LSU represented by the respective entry. For example, the entry 67 may reference a table 72, referred to herein as an “LSU table,” corresponding to the LSU represented by the entry 67. The LSU table 72 may include a header that contains information pertinent to the LSU as a whole. The LSU table 72 also may include entries 76-78 for separate contiguous logical data portions of the represented LSU; each such logical data portion corresponding to, and including a reference to, one or more contiguous physical locations (e.g., logical block address ranges) of a physical storage device (e.g., a cylinder and/or a group of tracks). In an embodiment disclosed herein, an LSU may contain any number of logical data portions depending upon how the LSU is initialized. However, in other embodiments, an LSU may contain a fixed number of logical data portions.
Each of the logical data portion entries 76-78 may correspond to a track table. For example, the entry 77 may correspond to a track table (or “LSU track table”) 82, which includes a header 84. The LSU track table 82 also includes entries 86-88, each entry representing an LSU track of the entry 77. In an embodiment disclosed herein, there are fifteen tracks for each contiguous logical data portion. However, for other embodiments, it may be possible to have different numbers of tracks for each of the logical data portions or even a variable number of tracks for each logical data portion. The information in each of the LSU track entries 86-88 may include a pointer (either direct or indirect—e.g., through another data structure) to a physical address of a physical storage device, for example, any of physical storage devices 24 of the storage system 20a (or a remote storage system if the system is so configured).
In addition to physical storage device addresses, or as an alternative thereto, each of the LSU track entries 86-88 may include a pointer (either direct or indirect—e.g., through another data structure) to one or more cache slots of a cache in the GM if the data of the logical track is currently in cache. For example, an LSU track entry 86-88 may point to one or more entries of cache slot table 300, described in more detail elsewhere herein. Thus, the LSU track table 82 may be used to map logical addresses of an LSU corresponding to the tables 62, 72, 82 to physical addresses within physical storage devices of a storage system and/or to cache slots within a cache.
In some embodiments, each entry 86-88 may specify a version of the data stored on the track, as described in more detail elsewhere herein.
Storage systems (e.g., the storage system 20a) also may maintain data structures (e.g., masking tables) that define I/O connectivity in terms of LSUs, storage ports and host ports; i.e., which ports of a host system (“host ports”; e.g., SCSI initiators) are permitted to perform I/O communications with which LSUs (e.g., identified with, and sometimes referred to as, a Logical Unit Numbers (LUNs)) over which ports of a storage system (“storage ports” e.g., SCSI targets). Defining (including initially defining and later modifying) which host ports are permitted to perform I/O communications with which LSUs over which storage ports, for example, using a masking table or other data structure, may be referred to as configuring or defining I/O connectivity between a host port, storage port and LSU, or more simply as “masking.”
The tables 62, 72, 72′, 82, 300 and 350 of
The system 100 may include a host system 102, switch 140 and data storage system 120. The host system 102 and data storage system 120 may communicate over one or more I/O paths through the switch 140. Elements 110a-110c denote connections between the host system 102 and switch 140. Element 112a-112c denote connections between the data storage system 120 and the switch 140. Element 130 may represent a physical storage device of the data storage system 120, such as, for example, a rotating disk drive or a solid state storage device (e.g., flash, storage class memory (SCM)), where the physical storage physical storage device 130 may be configured to include three LSUs—LUN5, LUN6 and LUN10. It should be noted that in the illustrative embodiment of
It should be appreciated that the descriptions provided in the following paragraphs may refer to particular examples using the switch 140 having a switching fabric for simplicity of illustration. Element 140 may be a single switch having a switching fabric, or a multi-switch having a multi-switch fabric and the like. Thus, element 140 may more generally denote a network having its own connectivity fabric or network fabric where the network may include one or more components providing the connectivity between the host system 102 and data storage system 120.
The host system 102 may be implemented as a server, and may include an application 104, a multi-path (MP) driver 106 and other components 108 such as, for example, one or more other device drivers and other code. An I/O request (specifying an I/O operation) from the application 104 may be communicated to the data storage system 120 using the MP driver 106 and one or more other components 108. The application 104 may be a database or other application which issues data operations, such as I/O operations, to the data storage system 120. Each of the I/O operations may be directed to a target device, such as one of the LSUs of physical storage device 130, configured to be accessible to the host system 102 over multiple I/O paths. As such, each of the I/O operations may be forwarded from the application 104 to the data storage system 120 over one of the possible multiple I/O paths.
The MP driver 106 may include functionality to perform any one or more different types of processing such as related to encryption, multi-pathing, mirroring, migration, and the like. For example, the MP driver 106 may include multi-pathing functionality for management and use of multiple I/O paths. For example, the MP driver 106 may perform I/O path selection to select one of the possible multiple I/O paths based on one or more criteria such as load balancing to distribute I/O requests for the target device across available active I/O paths. Load balancing may be performed to provide for better resource utilization and increased performance of the host system, data storage system, and network or other connection infrastructure. The MP driver 106 may be included in a commercially available product such as, for example, Dell EMC PowerPath® software made available by Dell EMC. Other components 108 of the host system 102 may include one or more other layers of software used in connection with communicating the I/O operation from the host system to the data storage system 120 such as, for example, Fibre Channel (FC) or SCSI drivers, a logical volume manager (LVM), or the like. The other components 108 may include software or other components used when sending an I/O operation from the application 104 to the data storage system 120, where such components may include those invoked in a call stack above and/or below the MP driver 106. For example, application 104 may issue an I/O operation which is communicated via a call stack including an LVM, the MP driver 106, and an FC or SCSI driver, e.g., as described elsewhere herein in more detail.
The data storage system 120 may include one or more physical storage devices, such as physical storage device 130, where each such physical storage device may be configured to store data of one or more LSUs. Each of the LSUs having data stored on the physical storage device 130 may be configured to be accessible to the host system 102 through one or more I/O paths. For example, all LSUs of physical storage device 130 may be accessible using ports of the three FEs 122a-122c, also denoted respectively as host adapters HA1, HA2 and HA3. The multiple I/O paths allow the application I/O requests to be routed over multiple I/O paths and, more generally, allow the LSUs of physical storage device 130 to be accessed over multiple I/O paths. In the event that there is a component failure in one of the multiple I/O paths, I/O requests from applications can be routed over other alternate I/O paths unaffected by the component failure. The MP driver 106 may be configured to perform load balancing in connection with I/O path selection, as well as other processing. The MP driver 106 may be aware of, and may monitor, all I/O paths between the host system and the LSUs of the physical storage device 130 in order to determine which of the multiple I/O paths are active or available at a point in time, which of the multiple I/O paths are unavailable for communications, and to use such information to select an I/O path for host system-data storage system communications.
In the example of the system 100, each of the LSUs of the physical storage device 130 may be configured to be accessible through three I/O paths. Each I/O path may be represented by two path endpoints having a first endpoint on the host system 102 and a second endpoint on the data storage system 120. The first endpoint may correspond to a port of a host system component, such as a host bus adapter (HBA) of the host system 102, and the second endpoint may correspond to a port of a data storage system component, such as a port of an HA of the data storage system 120. In the example of the system 100, elements A1, A2 and A3 each denote a port of a host system 102 (e.g., a port of an HBA), and elements B1, B2 and B3 each denote a port of an HA of the data storage system 120. Each of the LSUs of the physical storage device 130 may be accessible over three I/O paths—a first I/O path represented by A1-B1, a second I/O path represented by A2-B2 and a third I/O path represented by A3-B3.
Functionality for performing multi-pathing operations, such as may be performed by Dell EMC PowerPath software, may be included in one of the driver extension modules such as a multi-path extension module. As described above, the MP driver may perform processing in connection with multiple I/O path management and selecting one of a plurality of possible I/O paths for use in connection with processing I/O operations and communicating with the data storage system, such as data storage system 120 of
In some embodiments, layers 121-125c are implemented on a host (e.g., the host system 102) coupled to a data storage system (e.g., the data storage system 120) that is an intelligent data storage system having its own mapping layer 127 such that the LSU known or exposed to the host system may not directly correspond to a physical storage device such as a disk drive. In such embodiments, the LSU specified by the host system in the I/O operation may be further mapped by the data storage system using its mapping layer 127. For example, an LSU specified by the host system may be mapped by the data storage system to one or more physical drives, and multiple LSUs may be located on a same physical storage device, multiple physical drives, and the like.
The MP driver 106, as well as other components illustrated in
In operation, an application executing at application layer 121 may issue one or more I/O requests specifying I/O operations (e.g., read and write operations) to logical volumes (implemented by the LVM 125a) or files (implemented using the file system 123), whereby such I/O requests may be mapped to I/O communications (specifying the I/O operation) directed to LSUs of the data storage system. Such I/O operations from the application layer 121 may be directed to the MP driver 106 after passing through any intervening layers such as, for example, the layers 123 and 125a. Communications between an initiator port of the host system and a target port of a data storage system (e.g., target port of an HA) may include those related to I/O operations and other non-IO commands such as related to host system control operations. I/O operations may include, for example, read and write operations with respect to data stored on an LSU.
In connection with the SCSI standard, an I/O path may be defined between an initiator port of the host system and a target port of the data storage system. An I/O request may be sent from the host system (e.g., from a component thereof such as an HBA), which may be referred to as an initiator, originator or source with respect to the foregoing I/O path. The host system, as the initiator, sends I/O requests along the I/O path to a data storage system (e.g., a particular component thereof such as an HA having a port with a network address), which may be referred to as a target, destination, receiver, or responder. Each physical connection of an I/O path may be between a first endpoint which is a port of the host system (e.g., such as an HBA having ports such as denoted as A1-A3 of
In some embodiments of the invention, a host system is directly connected to an internal fabric of a storage system; i.e., the host is connected to the internal fabric without an intervening director (e.g., FA) or other component of the storage system controlling the host system's access to the internal fabric. For example, rather than a host system (e.g., host 14a) being physically coupled to a network (e.g., network 18), which is coupled to an FA (e.g., host adapter 21a), which is coupled to an internal fabric (e.g., internal fabric 30) of a storage system (e.g., storage system 20a), where the FA controls the host system's access to other components (e.g., GM 25b, other directors 37a-n) of the storage system over the internal fabric as illustrated in
In some embodiments, the GM may include persistent memory for which data stored thereon (including state information) persists (i.e., remains available) after the process or program that created the data terminates, perhaps even after the storage system fails (for at least some period of time). In some embodiments, the internal fabric exhibits low latency (e.g., when IB is employed). In such embodiments, by enabling a host system to directly access GM of the storage system, which may include persistent memory, host systems may be configured to expand their memory capacity, including persistent memory capacity by using the memory of the storage system. Thus, a system administrator could expand the memory capacity, including persistent memory capacity of the hosts of a storage network without having to purchase, deploy and configure new host systems. Rather, the system administrator may configure existing host systems to utilize the GM of the storage system, and/or purchase, install and configure one or more storage system interfaces (SSIs; described elsewhere herein in more detail) on existing host systems, which may result in significant savings in time and cost. Further, because of the security advantages provided by the SSI described in more detail elsewhere herein, use of the GM may prove more secure than memory, including persistent memory, added to host systems to expand memory capacity.
In some embodiments, an SSI located externally to the storage system may be provided that serves as an interface between the host system and storage system. The SSI may be part of the host system, and in some embodiments may be a separate and discrete component from the remainder of the host system, physically connected to the remainder of the host system by one or more buses that connect peripheral devices to the remainder of the host system. The SSI may be physically connected directly to the internal fabric. In some embodiments, the SSI may be implemented on a card or chipset physically connected to the remainder of a host system by a PCIe interconnect.
A potential benefit of implementing an SSI as a physically separate and discrete component from the remainder of a host system is that the SSI's resources may be configured such that its resources are not available for any functions, tasks, processing or the like on the host system other than for authorized I/O processing. Thus, I/O performance may be improved and more deterministic, as SSI resources may not be depleted for non-IO-related tasks on the host system. Further, as a physically separate and discrete component from the remainder of the host system, the SSI may not be subject to the same faults as the remainder of the system, i.e., it may be in a different fault zone from the remainder of the host system.
The SSI may provide functionality traditionally provided on storage systems, enabling at least some I/O processing to be offloaded from storage systems to SSIs, for example, on host systems. Metadata about the data stored on the storage system may be stored on the SSI, including any metadata described herein. The SSI may be configured to determine whether an I/O operation is a read or write operation, and process the I/O operation accordingly. If the I/O operation is a read operation, the SSI may be configured to determine from metadata whether the data to be read is in cache on the storage system. If the data is in cache, the SSI may read the data directly from cache over the internal fabric without use of CPU resources of a director, and, in some embodiments, without use of a director at all. If the data is not in cache, the SSI may determine, from the metadata, the physical storage device and physical location (e.g., address range) therein of the data to be read. The data then may be read from the physical storage device over the internal fabric without use of a director. Data may be read from a cache or physical storage device to the SSI using RDMA communications that do not involve use of any CPU resources on the storage system, SSI or the host system (e.g., other parts thereof), thereby preserving CPU resources on the storage network.
The I/O processing capabilities of an SSI may be used to offload I/O processing from a storage system, thereby reducing consumption of I/O compute resources on the storage system itself. The overall storage compute capacity of a storage network may be increased without having to upgrade or add a storage system.
In some embodiments, an SSI may implement one or more technology specifications and/or protocols, including but not limited to, NVMe, NVMeoF and IB. For example, SSI may be configured to exchange I/O communications with the remainder of the host system in accordance with NVMe. In embodiments in which an SSI is configured to communicate in accordance with NVMe, as opposed to in accordance with a native platform (including an OS or virtualization platform) of the host system, significant development and quality assurance costs may be realized, as developing or upgrading an SSI for each new or updated native platform may be avoided. Rather, the native platform may conform to NVMe, an industry standard, and support an OS-native inbox NVMe driver.
In some embodiments, secure access to data on a storage system via direct connection to an internal fabric may be provided. An SSI may validate each I/O communication originating on the host system before allowing a corresponding I/O communication to be transmitted on the internal fabric. The validation may include applying predefined rules and/or ensuring that the I/O communication conforms to one or more technologies, e.g., NVMe. Additional security measures may include requiring validation of any SSI software or firmware before loading it onto the SSI, for example, using digital signatures, digital certificates and/or other cryptographic schemes, to ensure unauthorized code is not loaded onto the SSI that could enable unauthorized I/O activity on a storage system. Further, in some embodiments, the SSI may be configured to encrypt I/O communications originating on a host system and to decrypt I/O communications received from the storage system, for example, in embodiments in which data is encrypted in flight between the host system to physical storage devices, and data may be encrypted at rest in memory of the storage system and/or on physical storage devices.
In addition, data integrity (e.g., checksums) in accordance with one or more technologies (e.g., T10DIF) may be employed by the SSI on I/O communications exchanged between host systems and data storage systems, by which end-to-end data integrity between a host system and physical storage devices may be implemented, as described in more detail herein.
In some embodiments, in addition to an SSI communicatively coupled between a host operating system and an internal fabric of a storage system, a storage network may include an interface communicatively coupled between an internal fabric and a DAE that encloses a plurality of physical storage devices; i.e., a fabric-DAE interface (“FDI”). The FDI may be configured to employ any of a plurality of technologies, including NVMe, NVMeoF and IB, as described in more detail herein. In such embodiments, I/O communications configured in accordance with NVMe may be implemented end-to-end from a host system to physical storage device, as described in more detail herein.
As described in more detail herein, through an SSI, a host system may exchange I/O communications, including control information (e.g., commands) and data, with GM including cache along an I/O path including internal fabric without use of compute resources of any of directors. Further, through an SSI, a host system may exchange I/O communications, including control information (e.g., commands) and data, with physical storage devices along an I/O path including internal fabric and not including use of directors. Thus, an I/O path in a known storage network, which may include an HBA, an external network, an FA, an internal fabric, a BE, a PCI switch and a physical storage device, may be replaced with an I/O path in accordance with embodiments of the invention, which includes an SSI, an internal fabric, an FDI and a physical storage device. These new I/O paths, eliminating use of external networks and director compute resources (or directors altogether) may produce reduced response times for certain I/O operations, as described in more detail elsewhere herein.
By removing an external network from the I/O path between a host system and a storage system, and routing I/O requests (e.g., all I/O requests on a storage network) through one or more SSIs, the possible sources of malicious actions or human error can be reduced; i.e., the attack surface of a storage system can be reduced. Further, by implementing validation logic as described in more detail herein, in particular as close as possible (logically) to where an SSI interfaces with a remainder of a host system (e.g., as close as possible to physical connections to peripheral device interconnects), for example, within an NVMe controller, the storage system may be made more secure than known storage networks having I/O paths including external networks. To further reduce access to an SSI, an NVMe driver may be configured as the only interface of an SSI made visible and accessible to applications on a host system. Any other interfaces to an SSI, for example, required for administration, may be made accessible only through certain privileged accounts, which may be protected using security credentials (e.g., encryption keys).
It should be appreciated that, although embodiments of the invention described herein are described in connection with use of NVMe, NVMeoF and D3 technologies, the invention is not so limited. Other technologies for exchanging I/O communications, for example, on an internal fabric of a storage system, may be used.
Storage network 600 may include any of: one or more host systems 14a-n (described in more detail elsewhere herein); network 18 (described in more detail elsewhere herein); one or more host systems 614a-n; one or more storage systems 620a-n; and other components. Storage system 620a may include any of: GM 640 (e.g., any of 25b, 220a-n and/or 220); one or more directors 637 (e.g., 37a-n); a plurality of physical storage devices 624 (e.g., 24), which may be enclosed in a disk array enclosure 627 (e.g., 27); internal fabric 630 (e.g., internal fabric 30); FDI 606, other components; or any suitable combination of the foregoing. Internal fabric 630 may include one or more switches and may be configured in accordance with one or more technologies, for example, D3. In some embodiments, at least a portion of GM 640, including at least a portion of cache 642, may reside on one or more physically discrete processing nodes (e.g., circuit boards) on which one of the directors 637 also resides, for example, in a manner similar to (or the same as) storage processing nodes 212a-n described in relation to
Each of host systems 614a-n may include SSI 616 connected directly to internal fabric 630 and configured to communicate with GM 640 and physical storage devices 624 (e.g., via FDI 606) over the internal fabric 630 independently of any of the directors 637 or any external network, for example, network 18. In embodiments in which one or more directors 637 may be included on one or more same processing nodes as at least a portion of GM 640, including at least a portion of cache 642 in some embodiments, SSI 616 may be configured to communicate with such GM 640, including cache 642, directly without use of any of the directors 637 or compute resources (e.g., processing cores of a CPU and/or CPU complex). For example, SSI 616 may be configured to use RDMA as described in more detail herein. Thus, embodiments of the invention in which a host system, or more particularly an SSI, communicates directly with a GM or cache of a storage system include: the host system communicating with a portion of GM or cache not included in any processing node (i.e., where GM is not included on a processing node, but is a physically separate component) independently of any director; and/or the host system communicating with a portion of GM or cache included in on a processing node that also includes one or more directors independently of any compute resources or any of the directors. In both cases, communicating directly with a GM or cache of a storage system does not involve use of any compute resources of the director.
As illustrated in
One or more of the directors 637 may serve as BEs (e.g., BEs 23a-n) and/or FAs (e.g., host adapter 21a-n), and enable I/O communications between the storage system 620a and hosts 14a-n and/or 614a-n over network 18, for example, as described in relation to
SSI 616 may be implemented as SSI 716 described in relation to
Host system 700 (e.g., one of host systems 614a-n) may include any of: operating system (OS) 701; an SSI 716 (e.g., SSI 616); one or more peripheral device interconnects 703; other components; and any suitable combination of the foregoing. Host OS 701 may be configured to execute applications running on the host system, which may result in I/O operations for data stored on any of storage systems 620a-n, requiring I/O communications to be exchanged between the host system and the one or more storage systems 620a-n. Host OS 701 may be any suitable operating system for processing I/O operations, for example, a version of Linux, or a hypervisor or kernel of a virtualization platform, for example, a version of VMware ESXi™ software available from VMware, Inc. of Palo Alto, Calif. Other operating systems and virtualization platforms that support an NVMe driver may be used.
In some embodiments, SSI 716 may be physically separate and discrete from the remainder of host system 700, the remainder including the OS 701 of the host system and the hardware and firmware on which the OS 701 executes, and SSI 716 may be pluggable into host system 700, which may be physically configured to receive SSI 716. In such embodiments, the SSI 716 may be considered a first physical part of the host system, for example, a peripheral component or device of the host system, and the remainder of the host system may be considered a second physical part of the host system. For example, SSI 716 may be configured to physically connect to the other part of the host system 700 by the one or more peripheral device interconnects 703, which may be configured in accordance with one or more technologies (e.g., PCIe, GenZ, another interconnect technology, or any suitable combination of the foregoing). An interconnect configured to connect to, and enable communications with, a peripheral component or device may be referred to herein as a “peripheral device interconnect,” and a peripheral device interconnect configured in accordance with PCIe referred to herein as a “PCIe interconnect.” SSI 716 may be implemented on a card or chipset, for example, in the form of a network interface controller (NIC), which may be configured with additional logic as described herein such that the resulting device may be considered a smart NIC (“SmartNIC”). As is described in more detail herein, SSI 716 may include an operating system for executing one or more I/O-related functions. Thus, in some embodiments, a first one or more operating systems (e.g., host OS 701) may be executing applications (e.g., on first part of the host 700) that result in I/O operations, while SSI 716 includes one or more second operating systems for performing functions and tasks on SSI 716 in relation to processing such I/O operations, such functions and tasks described in more detail elsewhere herein.
In some embodiments, SSI 716 may be configured to communicate according to a PCIe specification over one or more peripheral device interconnects 703, and SSI 716 may be configured to communicate according to an NVMe specification such that the SSI 716 presents itself as one or more NVMe devices (e.g., drives) to the host system 700. For example, the host interface 706 may include an NVMe controller 708 configured to exchange I/O communication according to NVMe with NVMe queues within an NVMe driver 702 of OS 701. That is, the OS 701 of the host system 700 may include an NVMe driver 702 configured to exchange I/O communications with the NVMe controller 708 in accordance with NVMe. To this end, the NVMe driver 702 may include at least two I/O queues, including one or more submission queues (SQs) 704a for submitting commands via a peripheral device interconnect 703 (configured as a PCIe interconnect) to NVMe controller 708, and may have one or more completion queues (CQs) 704b for receiving completed commands from NVMe controller 708 via one or more interconnects 703. Each SQ may have a corresponding CQ, and, in some embodiments, multiple SQs may correspond to the same CQ. In some embodiments, there may be up to 64K I/O queues in accordance with a version of the NVMe specification. The NVMe driver 702 also may include an admin SQ and CQ pair for control management in accordance with a version of the NVMe specification, and NVMe driver 702 and NVMe controller 708 may be configured to exchange control management communications with each other using admin SQs and CQs in accordance with a version of the NVMe specification.
SSI 716 may include any of: host interface 706; security logic 710; I/O processing logic 717; storage metadata (MD) 722; storage system communication interface (SSCI) 729; registration logic 727; memory 723; other components; or any suitable combination of the foregoing.
Registration logic 727 may be configured to register host system 700 and/or SSI 716 with storage system 620a when SSI 716 is connected to internal fabric 630, to enable future communication between the storage system 620a and internal fabric 630.
Security logic 710 may include any of: I/O validation logic 711; cryptographic logic 712; code validation logic 713; security credentials 714; other components; or any suitable combination of the foregoing. I/O validation logic 711 may prevent any undesired (e.g., invalid) communications from being further processed by SSI 716 or storage system 620a. Security logic 710, and more specifically I/O validation logic 711, may be a first component of SSI 716 to act on a communication received on one of the peripheral device interconnects 703, to ensure that any undesired communications do not proceed any further within SSI 716 and storage system 620a. To this end, it should be appreciated that one or more aspects of security logic 710, including I/O validation logic 711 and code validation logic 713, or portions thereof, may be implemented as part of host interface 706, for example, as part of NVMe controller 708.
IO validation logic 711 may include logic that verifies that a communication received on one of peripheral device interconnects 703 is indeed an I/O communication authorized to be transmitted on SSI 716. For example, I/O validation logic 711 may be configured to ensure that a received communication is an I/O communication properly configured in accordance with NVMe, and to reject (e.g., discard or drop) any received communications not properly configured. Further, I/O validation logic 711 may be configured to allow only a certain subset of I/O operations, for example, read or write operations, and reject other I/O operations, for example, operations to configure storage and/or other storage management operations. Such stipulations may be captured as one or more user-defined rules that may be defined and stored (e.g., in a rules data structure) within SSI 716. It should be appreciated that rules may be specific to one or more storage-related entities, for example, users, groups of users, applications, storage devices, groups of storage devices, or other property values. Thus I/O validation logic 711 may be configured to implement any of a variety of business rules to control access to resources on storage system 620a.
Cryptographic logic 712 may be configured to encrypt data included in I/O communications received from host OS 701 and before repackaging the data (in encrypted form) in I/O communications transmitted over internal fabric 630 to components of storage system 620a. Cryptographic logic 712 also may be configured to decrypt data from I/O communications received from internal fabric 620a before sending the unencrypted data in I/O communication to host OS 701. Any of a variety of cryptographic schemes may be used, including use of symmetric and/or asymmetric keys, which may be shared or exchanged between SSI 716 of the host system, one of more storage systems 620a-n, and one or more SSIs of other host systems 614a-n, depending on what entities are entitled access to the data. For example, during a manufacturing and/or configuring of SSIs 716 and/or storage systems 620a-n, one or more encryption keys and/or other secrets (collectively, “security credentials”) may be shared, to enable implementation of the given cryptographic scheme, and may be stored as part of security credentials 714.
In embodiments in which data is encrypted on SSI 716 before being transmitted to the storage system 620a, the data may be stored in encrypted form in physical storage devices 624 and/or GM 640. In such embodiments, directors 637 and other components that may be authorized to access the encrypted data also may be configured to implement whatever cryptographic scheme is being employed, which may be desirable for host systems (e.g., host systems 14a-n) that may access storage system 620a by means other than an SSI as described herein. In some known storage systems, physical storage devices may be self-encrypting drives that encrypt data received from BEs, and then decrypt the data when it is retrieved for BEs. This may be considered a form of data-at-rest encryption. In embodiments of the invention in which data is encrypted on SSI 716, and transmitted to physical storage devices 624 in encrypted form to be stored, it may be desirable that physical storage devices 624 do not employ their own encryption, as the data will arrive encrypted. That is, encrypting the already-encrypted data would be redundant, and a waste of processing resources. Further, self-encrypting drives may be more expensive than drives not including this feature. Thus, if there is no need for physical storage devices 624 to encrypt and decrypt data, physical storage device not having self-encryption, but otherwise having the same or similar capabilities, may be acquired at reduced cost.
By encrypting data on a host system, e.g., as part of an SSI 716, data may not only be able to be encrypted while at rest, but also while in transit. That is, in embodiments of the invention, data may be encrypted in transit on an I/O path from a host system to a physical storage device (i.e., end-to-end) as well as being encrypted at rest on a physical storage device or in memory (e.g., cache) of a storage system.
As described in more detail elsewhere herein, SSI 716 may be implemented in various combinations of hardware, software and firmware, including microcode. In some embodiments of SSI 716 implemented using software and/or firmware, the software and/or firmware, and updates thereto, may be subject to verification of digital signature before being allowed to be installed on SSI 716. For example, the security credentials 714 may include a public certificate that includes a cryptographic key (e.g., a public key of a PKI pair or the like), which may be embedded within the software and/or firmware initially installed on SSI 716 (e.g., at the manufacturer of SSI 716). The public certificate also may specify a validity period for the public certificate. Each subsequent update of the software and/or firmware may be digitally signed with a digital signature based on an encryption scheme (e.g., PKI) involving the public key.
When a purported software and/or firmware update is received at SSI 716 including a digital signature, code validation logic 713 may use the public key (and the validity period) in the public certificate to validate the digital signature and thereby verify the authenticity of the update, for example, by exchanging communications with a certification service or the like of the SSI 716 manufacturer or a trusted third-party, using known techniques. The security credentials 714, including the public certificate and perhaps other credentials, and credentials used for encrypting and decrypting data, may be embedded within the software and/or firmware on the SSI 716 so that they are not accessible by the host system 700 or any other entity connected to the SSI 716. For example, the security credentials 714 may be stored within a trusted platform module (TPM) or the like within SSI 716. If the code validation logic determines the software or firmware update to be invalid, the update may not be installed on SSI 716. Such verification of the software and/or firmware may prevent an attacker from replacing software and/or firmware on SSI 716 with code that would allow access to resources within storage system 620a.
Storage metadata 722 may include any metadata about data stored on storage system 620a, including but not limited to any of the metadata described herein. For example, storage MD 722 may include any of master device table 762, LSU table 772, thin device table 772′, track table 782 and cache slot table 750, corresponding to master device table 62, LSU table 72, thin device table 72′, track table 82 and cache slot table 300, respectively. For example, each of tables 762, 772, 772′, 782 and 750 may include at least a portion of the metadata stored in 762, 772, 772′, 782 and 750, respectively; e.g., metadata corresponding to physical storage devices 624, and logical storage devices associated therewith, being used for applications running on host system 700. Use of such metadata is described in more detail elsewhere herein.
IO processing logic 717 may include one or more components for performing I/O operations in conjunction with storage system 620a. In some embodiments, one or more of these components embody I/O functionality, including data services, that is implemented on known storage systems. By implementing such I/O functionality on SSI 716 instead of on the storage system 620a, less storage system resources may be consumed, and overall I/O performance on the storage system may be improved. I/O processing logic 717 may include any of: device mapping logic 718; I/O path logic 720; integrity logic 721; messaging logic 724; RDMA logic 725; atomic logic 726; back-end logic 728; other components; or any suitable combination of the foregoing.
Device mapping logic 718 may be configured to map logical addresses of logical storage devices to locations (i.e., physical addresses) within physical storage devices using, e.g., any one or more of tables 762, 772, 772′ and 782, 750.
Device mapping logic 718 may be configured to determine, from the retrieved metadata, the location(s) within cache 642 and/or within one or more physical storage devices 624 corresponding to the logical location specified in the I/O operation. The I/O operation (e.g., read or write) then may be performed with respect to the determined cache location and/or one or more physical storage device locations. In some embodiments, the device mapping logic 718, in coordination with one or more other components of I/O processing logic 717, SSI 716 and/or host system 700, may perform aspects of I/O operations, including exchanging communications with components of storage system 620a over internal fabric 630, as described in U.S. patent application Ser. No. 16/389,383, titled “Host System Directly Connected to Internal Switching Fabric of Storage System,” to Wigmore et al., filed Apr. 19, 2019 (“Wigmore”), the entire contents of which are hereby incorporated by reference.
It should be appreciated that, while in some embodiments described herein, the metadata corresponding to a logical location may be used to determine a corresponding one or more locations in cache and/or a physical storage location, the metadata may include a variety of information other than location information, as described in more detail elsewhere herein, and may be used for any of a variety of other purposes, for example, by one or more components of host system 700 and/or storage system 620a.
IO path logic 720 may be configured to determine what I/O path within storage system 620a to use to process an I/O operation. I/O path logic 720 may be configured to determine what path to take for an I/O operation based on any of a variety of factors, including but not limited to whether the I/O is a read or write; how complicated a state of the storage system is at the time the I/O operation is being processed; whether the data specified by the I/O operation is in a cache of the storage system; other factors; or a combination of the foregoing. For example, based on one or more of the foregoing factors, I/O path logic 720 may determine whether to process an I/O request by: sending a communication to a director; directly accessing a cache on the storage system (i.e., without using any compute resources of a director) or accessing a physical storage device without using a director (e.g., via an FDI). I/O path logic 720 may be configured to determine what I/O path within storage system 620a to use to process an I/O operation as described in more detail in Wigmore.
Integrity logic 721 may be configured to implement one or more data integrity techniques for I/O operations. Some data storage systems may be configured to implement one or more data integrity techniques to ensure the integrity of data stored on the storage system on behalf of one or more host systems. One such data integrity technique is called DIF (data integrity field), or “T10DIF” in reference to the T10 subcommittee of the International Committee for Information Technology Standards that proposed the technique. Some storage systems, for example, in accordance with one or more technology standards, store data arranged as atomic storage units called “disk sectors” having a length of 512 bytes. T10DIF adds an additional 8 bytes encoding a checksum of the data represented by the remaining 512 byes, resulting in data actually being stored as 520-byte atomic units, including 512 bytes of data and 8 bytes of checksum data in accordance with T10DIF. In embodiments of the invention in which storage system 620a is implementing T10DIF, integrity logic 721 may be configured to implement T10DIF, thereby converting 512-byte units of data in I/O communications received from host OS 701 to 520-byte units of data in accordance with T10DIF to be transmitted in I/O communications to storage system 620a. In such embodiments, integrity logic 721 also may be configured to convert 520-byte units of data in I/O communications received from storage system 620a to 512-byte units of data to be transmitted in I/O communications to host OS 701. In such embodiments, data integrity on a storage network (e.g., storage network 600) may be improved by implementing T10DIF on an I/O path from a host system to a physical storage device (e.g., end-to-end).
Processing I/O operations in accordance with embodiments of the invention may include exchanging RDMA communications, control (e.g., command) communications and atomic communications between host system 700 and storage system 620a. RDMA logic 725, messaging logic 724, and atomic logic 726, respectively, may be configured to implement such communications. Atomic communications involve performing exclusive locking operations on memory locations (e.g., at which one or more data structures described herein reside) from which data is being accessed, to ensure that no other entity (e.g., a director) can write to the memory location with other data. The exclusive locking operation associated with an atomic operation introduces a certain amount of overhead, which may be undesired in situations in which speed is of greater performance.
It may be desirable for host system 700; e.g., SSI 716, to know information (e.g., a state) of one or more physical storage devices 624, for example, whether a physical storage device is off-line or otherwise unavailable, e.g., because of garbage collection. To this end, in some embodiments, back-end logic 728 may monitor the status of one or more physical storage devices 624, for example, by exchanging communications with FDI 606 over internal fabric 630.
The SSCI 729 may include logic for steering and routing I/O communications to one or more ports 731 of SSI 716 physically connected to internal fabric 630, and may include logic implementing lower-level processing (e.g., at the transport, data link and physical layer) of I/O communications, including RDMA, messaging and atomic communications. In some embodiments of the invention, communications between SSI 716 and components of storage system 620a (e.g., directors 637, GM 640 and FDI 606) over internal fabric 630 may be encapsulated as NVMeoF command capsules in accordance with an NVMeoF specification. For example, SSCI 729 may include logic for encapsulating I/O communications, including RDMA, messaging and atomic communications, in accordance with NVMeoF. Thus, in some embodiments, I/O communications received from NVMe driver 702, configured in accordance with NVMe, may be converted to NVMeoF command capsule communications for transmission over the internal fabric 630. SSCI 729 also may include logic for de-capsulating NVMeoF command capsules, for example, into NVMe communications to be processed by I/O processing logic 717.
The SSCI 729 (and components of the storage system 620a interfacing with the internal fabric 630) may be configured to address communication to other components; e.g., GM 640, FDI 606, directors 637, in accordance with one or more technologies being used to communicate over internal fabric 630. For example, in embodiments in which D3 is employed to communicate over internal fabric 630, SSCI 729 may be configured to address communication to other components using IB queue pairs. Aspects of SSCI 729 may be implemented using a network adapter (e.g., card or chip), for example, a ConnectX®-5 or ConnectX®-6 dual-port network adapter available from Mellanox Technologies, Ltd. of Sunnyvale, Calif. (“Mellanox”), for example, as part of a SmartNIC.
The SSI 716 may be implemented as a combination of software, firmware and/or hardware. For example, SSI 716 may include certain hardware and/or firmware, including, for example, any combination of printed circuit board (PCB), FPGA, ASIC, or the like, that are hardwired to perform certain functionality, and may include one or more microprocessors, microcontrollers or the like that are programmable using software and/or firmware (e.g., microcode). Any suitable microprocessor may be used, for example, a microprocessor including a complex instruction set computing (CISC) architecture, e.g., an x86 processor, or processor having a reduced instruction set computing (RISC) architecture, for example, an ARM processor. SSI 716 may be part of a microprocessor or separate therefrom. In embodiments in which a microprocessor is employed, any suitable OS may be used to operate the microprocessor, including, for example, a Linux operating system. In some embodiments, the combination of software, hardware and/or firmware may constitute a system-on-chip (SOC) or system-on-module (SOM) on which SSI 716 may be implemented, e.g., as part of a SmartNIC. For example, in some embodiments, SSI 716 may be implemented, at least in part, using a BlueField™ Multicore System On a Chip (SOC) for NVMe storage, available from Mellanox, which may be further configured with logic and functionality described herein to constitute a SmartNIC.
Returning to
The FDI 606 may be implemented as a combination of software, firmware and/or hardware including, for example, any combination of a printed circuit board (PCB), FPGA, ASIC, or the like, that are hardwired to perform certain functionality, and may include one or more microprocessors, microcontrollers or the like that are programmable using software and/or firmware (e.g., microcode). Any suitable microprocessor may be used, for example, a microprocessor including a complex instruction set computing (CISC) architecture, e.g., an x86 processor, or processor having a reduced instruction set computing (RISC) architecture, for example, an ARM processor. In some embodiments, the combination of software, hardware and/or firmware may constitute a system-on-chip (SOC) or system-on-module (SOM) on which FDI 606 may be implemented. For example, in some embodiments, the FDI 606 may be implemented using a BlueField™ Multicore SOC for NVMe storage, available from Mellanox.
The systems and data structures described in relation to
The system 800 may include one or more host systems 802a-n and a storage system 820. The host system 802a may be, include, be included within and/or be a variation of the host system 700 and/or one of the host systems 14a-n. The host system 802a may include an OS 806 and one or more SSIs, including SSI 816, which may be, include, be included within and/or be a variation of the SSI 716. The SSI 816 may be directly connected to the internal fabric 830 (e.g., 630) of the storage system 820, and be configured to directly communicate with global memory (GM) 825 and other components of the storage system 820 without use of any directors and/or other compute resources of the storage system 820. The SSI 816 may be configured to communicate directly with the GM 825 and other components of the storage system 820 using RDMA technology.
The storage system 820 may be, include, be included within and/or be a variation of the storage system 620a. The storage system 820 may include one or more storage processing nodes 840a-n, each of which may be, include, be included within and/or be a variation of a storage processing node 212a-n. The storage processing node 840a may include any of: a CPU 842, dynamic SCM logic (DCL) 846, a page allocation queue (PAQ) 847; a translation lookaside buffer (TLB) 847; a GM segment 827 of GM 825; SCM 860; one or more other components; or any suitable combination of the foregoing.
The SCM 860 may include two primary parts: a memory part designated for use as memory (i.e., for use in a memory context) and a storage part designated for use as storage (i.e., for use in a storage context). For example, in some embodiments of the invention in which Intel Optane memory is employed, a memory part may be a part of SCM designated as being in volatile memory mode, and a storage part may be a part of SCM designated as application direct storage. Each of the storage part and the memory part are not necessarily comprised of physically contiguous portions of the SCM 860, but rather may include one or more non-physically contiguous portions. For example, the SCM 860 may include storage portions 862a,d and h, which are not physically contiguous as illustrated in
While the SCM 860 is illustrated as being part of GM segment 827, it should be appreciated that only the memory portions of SCM 360 are actually part of the GM segment 827 and GM 825. That is, logic on the storage system 820 that manages GM 825, for example, the memory management module 32, may be configured to recognize memory portions 862b,c,e,f and g and 863a,b,e,f and h as being part of the GM 825, but not the storage sub-portions.
The SCM 860 may be implemented as one or more physically discrete components, for example, multiple chips of a DIMM of the storage system. Thus, two or more of the portions of the SCM 860, whether designated as memory or storage, may reside on separate discrete components an SCM.
The storage processing node 840n may include the same and/or similar logic and/or data structures as the storage processing node 840a. For example, the storage processing node may include, among other components, a GM segment 828, which may include an SCM 861. The SCM 861 may include a memory part including portions 863a, b, e, f and h and a storage part including storage portions 863c, d and g. Similar to the SCM 860, the SCM 861 may be implemented as one or more physically discrete components, for example, multiple chips of a DIMM of the storage system. Thus, the SCM available for use as memory or storage on a storage system may include multiple physically discrete SCM components (e.g., DIMM chips) spanning multiple storage processing nodes (e.g., director boards of PowerMax) of a storage system (e.g., 820).
In some embodiments of the invention, each of portions 862a-h and/or portions 863a-h may be configured to have a uniform size such as, for example, a certain number of Gigabytes. As described in more detail herein, in some embodiments, a certain number or percentage (e.g., all) of the portions of an SCM may initially be designated as memory until a need for storage arises (e.g., in response to one or more write requests), at which time one or more portions of the SCM may be converted for use as storage.
In some embodiments of the invention, a thing provisioning technique is employed with respect to provisioning SCM (e.g., the SCM 860 and the SCM 861) for storage. That is, while a certain amount of SCM may be provisioned (e.g., to a host application and/or LSU), SCM storage may not be actually allocated to the host application and/or LSU until needed—i.e., until more SCM storage is needed in response to a write operation. For example, in response to a write request that specifies a VML corresponding to a host application and/or LSU, if the write request is a first write request to the VML, one or more SCM storage pages may be allocated to the VML.
The CPU 842 may include a virtual memory mapping table (VMMT). The VMMT may include a plurality of entries, each entry representing and specifying a virtual memory location (e.g., address), and specifying the virtual memory location. For each entry, if a physical memory location (e.g., physical address of storage sub-portion) has been allocated to the virtual memory location, the entry may specify the physical address of the storage sub-portion (i.e., “page”). The size of each page may be predefined, for example, 4 KB.
In response to receiving a read operation from one of the host systems 802a-n, the CPU may determine the page location (i.e., address) allocated to the virtual memory location specified in the read operation by accessing the entry in the VMMT corresponding to virtual memory location. In response to receiving a write operation from one of the host systems 802a-n, the CPU may consult the VMMT to determine if there is an entry in the VMMT for the virtual memory location specified in the write operation, and, if there is such an entry, return the page location specified by the entry. If there is not an entry in the VMMT for the virtual memory location, e.g., if the write is a first write to the virtual memory location, the CPU may be configured (e.g., with a memory management unit (MMU)) to determine a page for the virtual memory location (i.e., translate a VML to a page address), allocate the page to the virtual memory location and create an entry in the VMMT mapping the virtual memory location to the page location.
Accessing the CPU each time an I/O operation is received from a host system to determine a page location consumes CPU resources. Accordingly, in some embodiments of the invention, the storage processing node 840a includes the TLB 847. The TLB 847 may include at least some (e.g., all) of the information (e.g., one or more entries) of the VMMT 842. For example, the TLB 847 may include a plurality of entries 848, each entry representing a virtual memory location (e.g., address), and specifying the virtual memory location in a column 849. For each entry, if a page has been allocated to the virtual memory location, the entry may specify the page location (e.g., address) in a column 851. For example: an entry 848a may specify the page 868a of the storage portion 862d; an entry 848b may specify the page 868b of the storage portion 863c; and an entry 848n may specify the page 868c of the storage portion 862h.
The PAQ 849 may include a plurality of entries (i.e., slots), where each slot is either empty or specifies an SCM page, for example, by specifying the page's location. Each time a page is required to be allocated to a virtual memory location (e.g., to satisfy a first write request to a virtual memory location), the next slot in the PAQ is accessed to determine the page location, which then is allocated to the virtual memory location, where the allocation may be recorded in the VMMT 844 and the TLB 847, as described in more detail elsewhere herein. The PAQ 849 may be populated with SCM pages from the SCM storage pool, as described in more detail elsewhere herein.
The DCL 846 may be configured to dynamically convert (e.g., while the GM 825, including the SCMs 860 and 861, is on-line and in use for I/O processing) memory sub-portions of GM 825 to storage sub-portions, as described in more detail elsewhere herein. The DSL 846 may be configured to receive I/O requests from host systems, for example, from SSI 816 of the host system 802a. To conserve CPU resources and decrease I/O response time, the DCL 846 may be configured to access the TLB 847 to map a virtual memory address of an I/O operation to a physical location, e.g., one or more storage sub-portions of SCM 860 or 861, as described in more detail elsewhere herein. Further, the DSL 846 may include logic configured to populate the slots of the PAQ 849 with pages from the SCM storage pool, and to access such slots when a page allocation is needed, as described in more detail elsewhere herein. The DSL 846 may further include logic for provisioning portions of SCM available on the storage system (e.g., the SCM 860 and the SCM 861) as storage for applications and/or LSUs. The DSL 846 may include one or more discrete and/or integrated logical components that, when executed, perform methods described herein, steps thereof or portions of such methods or steps. Such components may be implemented as software, firmware, hardware, or any suitable combination thereof.
As described elsewhere herein, the host system 802a (and other host systems, e.g., the host system 802n) may be connected (e.g., via SSI 816) directly to the DSL 846 through the internal fabric 830. To this end, a logical connection may be defined between a host and the DSL 846, which may be referred to herein as a virtual protocol interconnect (VSI). An example of a VPI, VPI 835, is visually depicted in
A user interface (e.g., a graphical user interface (GUI) or command line interface (CLI)) may be provided by provisioning logic within the DSL 846, SSI 816, an application running on the host 802a, or any suitable combination of the foregoing, that enables users to configure GM, including provisioning the SCM 860 and the SCM 861. Such a provisioning interface may enable a user to provision portions of the SCM for use as storage or memory. The user may be enabled to provision SCM storage by specifying: a VSI corresponding to the SCM; a requested size; and whether the SCM should be provisioned for memory or storage. The provisioning logic may be configured to designate SCM portions accordingly for use as memory or storage. For example, in an embodiment in which the SCM is Intel Optane memory, the provisioning of the SCM may be referred to as memory registration. The provisioning logic may be configured to enable a user to specify any of a variety of other configuration parameter values, such as specifying which application, hosts and/or other entities have which permissions (e.g., read-only, read and write, etc.) with respect to the provisioned SCM.
In a step 902, SCM (e.g., 860 and/or 861) may be initially configured. For example, an administrator or other user may configure a variety of parameter values, e.g., using a GUI or CLI as described elsewhere herein. Some configuration parameters may include an amount of SCM to provision for memory and amount to provision as storage, and which hosts have what kinds of access (e.g., read-write, read only, etc.) to the SCM memory and/or storage. In some embodiments, a certain amount (e.g., all) of the SCM is initially configured by default as memory, and SCM memory is only converted dynamically to storage as needs to accommodate data storage.
In a step 904, SCM portions may be queued for allocation, for example, as described in relation to
In a step 908, it may be determined whether the virtual memory location specified by the write request is already mapped to a page of an SCM; i.e., whether an SCM page has already been allocated to the virtual memory location. This determination may be made, for example, by the DSL 846 determining if there is an entry in the TLB 847 for the VML, and if so what is the SCM page identified (e.g., by page address) by the entry.
If it is determined in the step 908 that an SCM page is already mapped to the VML, the data specified by the write request may be written to the SCM page in a step 916. If it is determined in the step 908 that an SCM page is not already mapped to the VML, then, in a step 910, it may be determined whether there is enough SCM storage (i.e., whether there are enough unallocated pages of SCM) to accommodate the data of the write request. For example, the step 910 may be performed by the DSL 846. If it is determined in the step 910 that there is enough such SCM storage, then, in a step 914, the next available SCM page (e.g., as specified in the PAQ 849) may be allocated to the VML.
The step 914 may be performed by the CPU 842, for example, under direction of the DSL 846. For example, the CPU 842 may include an MMU to obtain a next available SCM page from the PAQ 849. The CPU 842 (e.g., an MMU thereof) may be configured to create or update an entry in the VMMT 844 for the VML specified in the write request with an identifier (e.g., physical address) of the allocated page, and to update the TLB 847 with the same new or updated entry, in a step 915. The method 900 then may proceed to the step 916.
If it is determined in the step 910 that there is not enough such SCM storage, then, in a step 912, one or more SCM memory portions may be converted to SCM storage portions, for example, as described in relation to the method 1100 in connection with
In a step 1002, it may be determined whether there is a next empty slot in the PAQ; i.e., whether there is a next slot not already specifying an SCM page. If it is determined in the step 1002 that there is not a next empty slot, then the method 1000 may end or return to step 1002, for example, after a predefined delay, at a prescheduled time or in response to an event (e.g., user input).
If it is determined in the step 1002 that there is a next empty slot in the PAQ, then in a step 1004 it may be determined whether there is an SCM page available from the SCM storage pool to be queued; i.e., whether there is at least one SCM page currently designated for use as storage that is not already allocated to another VML. If it is determined in the step 1004 that there is an SCM page available to be queued, then, in a step 1008, the next slot in the PAQ may be updated to specify (e.g., point to) the SCM page.
If it is determined in the step 1004 that there is not an SCM page available to be queued, then, in a step 1006, one or more SCM memory portions may be converted to SCM storage portions, for example, as described in relation to the method 1100 in connection with
After performance of the step 1006, the method 1000 may proceed to the step 1008. After the performance of the step 1008, the method 1000 may return to the step 1002. In some embodiments of the invention, the steps 1002-1008 are iteratively performed until all slots of the PAQ specify an SCM page.
In a step 1102, a next SCM memory portion may be selected for conversion to SCM storage. For example, a queue of SCM memory portions may be maintained, and the next SCM memory portion on the SCM memory queue selected; or some other technique may be used to select a next SCM memory portion for conversion.
In a step 1104, the sub-portions of the selected SCM portion may be removed from one or more memory allocation mechanisms of the storage system. For example, one or more components of the storage system (e.g., the cache management module 34) may be configured to allocate memory sub-portions (e.g., pages) of GM, including SCM memory sub-portions, for use as cache, for example, for I/O operations to/from a host. In such embodiments, the step 1104 may remove such allocated sub-portions from the list, queue or other data structure that specifies GM sub-portions eligible for cache allocation.
In a step 1106, it may be determined whether there is data associated with any memory sub-portion of the selected SCM memory portion that needs to be processed. For example, if any data is currently cached in a memory sub-portion of the memory portion, the data may need to be evicted or de-staged (e.g., flushed to disk) before the memory portion is designated for use as storage.
If it is determined in the step 1106 that that there is no data associated with a memory sub-portion of the selected SCM memory portion that needs to be processed, then, in a step 1110, the SCM portion may be designated as storage; i.e., a designation of the SCM portion may be changed from memory to storage.
If it is determined in the step 1106 that that there is data associated with a memory sub-portion of the selected SCM memory portion that needs to be processed, then, in a step 1108, the data may be processed. For example, the data may be de-staged or evicted, data structures updated accordingly, and data cleared from the memory portion. After performance of the step 1108, the method 1000 may proceed to the step 1110.
The method 1200 may include traversing allocated SCM pages (e.g., all allocated SCM pages), determining whether they are being used, and reclaiming (e.g., de-allocating) inactive allocated SCM pages so they are eligible to be allocated again. In some embodiments, the method 1200 may start with a first allocated page and proceed through the last allocated page. The allocated pages may be ordered in any of a variety of ways, for example, according to SCM chip and/or physical address thereon. In some embodiments, the method 1200 may be performed continuously, iteratively traversing the allocated pages. In other embodiments, the method 1200 may be performed at a scheduled time (e.g., periodically) or in response to an event (e.g., user input). The method 1200 may be performed as a background process over which I/O processing takes precedence.
In a step 1202, it may be determined whether there is a next allocated SCM page, for example, in accordance with a predefined order. For example, on a first pass, the next allocated SCM page may be a first allocated page in a predefined order. If it is determined that there is a next allocated SCM page, then, in a step 1206, it may be determined whether there is any data stored in the SCM page. If there is data stored in the SCM page, this means that the allocated SCM page is still being used (i.e., is active) in which case the page may not be reclaimed and the method 1200 may return to the step 1202.
If it is determined in the step 1206 that there is no data stored in the allocated SCM page (e.g., the page consists of all “0”s), then, in a step 1208, any entries for the SCM page in the VMMT of the CPU and in the TLB may be cleared, and, in a step 1208, the page may be returned to the SCM storage pool so it may be allocated again for use.
If it is determined in the step 1202 that there is not a next allocated SCM page, for example, if the order has been fully traversed, then the method may end and be performed again later at a scheduled time or in response to an event. In some embodiments, for example, if the system is configured so that the method 1200 is performed continuously, the next allocated page may be set to the first allocated SCM page, and steps 1202-1210 repeated.
In some embodiments, it may be desired to ensure (to the extent feasible) that there is enough SCM storage available to be allocated in the event one or more large write requests and/or a burst of write requests are received.
The method 1300 may be performed continuously or may be performed at a scheduled time (e.g., periodically) or in response to an event (e.g., user input). The method 1300 may be performed as a background process over which I/O processing takes precedence.
In a step 1302, it may be determined whether there is enough SCM storage, i.e., whether there is enough SCM designated as storage. For example, a threshold amount of unallocated SCM storage may be defined, for example, based on past usage history and/or I/O forecasting. If the current amount of unallocated SCM storage is less than the predefined threshold amount, then SCM memory may be converted to SCM storage in a step 1304, as described in more detail elsewhere herein. After performance of the step 1304, the method 1300 may return to the step 1302.
If it is determined in the step 1302 that the current amount of unallocated SCM storage is equal to or greater than the predefined threshold amount, then the method 1300 may end and be performed again later at a scheduled time or in response to an event. If the system is configured so that the method 1300 is performed continuously, the method may perform the step 1302 again.
Various embodiments of the invention may be combined with each other in appropriate combinations. Additionally, in some instances, the order of steps in the flowcharts, flow diagrams and/or described flow processing may be modified, where appropriate. It should be appreciated that any of the methods described herein, including methods 900-1300, or parts thereof, may be implemented using one or more of the systems and/or data structures described in relation to
Software implementations of embodiments of the invention may include executable code that is stored one or more computer-readable media and executed by one or more processors. Each of the computer-readable media may be non-transitory and include a computer hard drive, ROM, RAM, flash memory, portable computer storage media such as a CD-ROM, a DVD-ROM, a flash drive, an SD card and/or other drive with, for example, a universal serial bus (USB) interface, and/or any other appropriate tangible or non-transitory computer-readable medium or computer memory on which executable code may be stored and executed by a processor. Embodiments of the invention may be used in connection with any appropriate OS.
As used herein, an element or operation recited in the singular and preceded with the word “a” or “an” should be understood as not excluding plural elements or operations, unless such exclusion is explicitly recited. References to “one” embodiment or implementation of the present disclosure are not intended to be interpreted as excluding the existence of additional embodiments that also incorporate the recited features. Furthermore, a description or recitation in the general form of “at least one of [a], [b] or [c],” or equivalent thereof, should be generally construed to include [a] alone, [b] alone, [c] alone, or any combination of [a], [b] and [c]. In addition, use of a an ordinal term, e.g., “first,” “second” or the like, to qualify a term for an item having multiple instances of the same name does not necessarily indicated a priority, precedence or temporal order between the instances unless otherwise indicated, but rather such ordinal terms may be used merely to distinguish between the separate instances.
Other embodiments of the invention will be apparent to those skilled in the art from a consideration of the specification or practice of the invention disclosed herein. It is intended that the specification and examples be considered as exemplary only, with the true scope and spirit of the invention being indicated by the following claims.