Backend storage testing method, device and medium based on SPDK
Based on the SPDK, target services and back-end storage are created for NVMe over Fabric, iSCSI and Ceph scenarios, setting up operation interfaces and loading NVMe drivers, the complexity and platform dependency problems of SPDK testing methods are solved, and multiple implementations are implemented. High performance and reliability testing in scenarios meets the needs of different storage systems.
Patent Information
- Application Number
- CN202311861552.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-12-29
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2043-12-29
AI Technical Summary
The existing back-end storage testing method based on SPDK has complexity, platform dependence, not suitable for all application scenarios, high customization requirements and not suitable for all storage hardware, making it difficult to cross-platform migration and meet in actual applications. Various storage requirements.
Provide a backend storage testing method based on SPDK, including creating target services and backend storage on the Target side, setting up the SoC side operation interface, loading the NVMe driver on the Host side, and performing performance, reliability, cluster configuration and data on the NVMe disk. Integrity testing, suitable for NVMe over Fabric, iSCSI and Ceph scenarios.
Through SPDK, high throughput, low latency and high reliability data transmission tests for different backend storage are realized, which improves the stability and performance of the storage system in actual scenarios and meets a variety of business needs.
Smart Images

Figure CN117667552B_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the field of computer technology, and in particular to a backend storage testing method, device and medium based on SPDK. Background Art
[0002] NVMe SSD is an application software acceleration library used as backend storage. Its core is the user-mode, asynchronous, polling NVMe (NVM Express, non-volatile memory host controller interface specification) driver. In order to give full play to the performance of the backend NVMe SSD, Intel developed SPDK (Storage Performance Development Kit). However, there are several problems in the test method based on SPDK backend storage: (1) Complexity: The use of SPDK is somewhat complex. For people who are not familiar with these fields, it is difficult to learn and use SPDK. (2) Platform dependence: The performance advantages of SPDK are usually fully utilized on specific hardware and operating system platforms, so cross-platform porting is difficult. (3) Not suitable for all application scenarios: Although SPDK performs well in some high-performance storage applications, it is not used in all application scenarios. For some smaller-scale applications that do not require extremely high performance, SPDK is too complex and is not the best choice. (4) Customization requirements: The powerful performance of SPDK comes from people who have a certain basic background in SPDK, and it takes a lot of time and effort to meet specific needs. (5) Not suitable for all storage hardware: Although SPDK supports a variety of storage devices, not all hardware can fully utilize their performance potential. Some older models or lower-end hardware may not be able to obtain similar advantages as high-performance hardware. Summary of the invention
[0003] In order to solve the above technical problems, the present disclosure provides a backend storage testing method, device and medium based on SPDK.
[0004] According to one aspect of the present disclosure, a backend storage testing method based on SPDK is provided, the method comprising:
[0005] Create a target service and a first backend storage corresponding to the target backend storage scenario on the Target side;
[0006] Adding the file name stored in the first back-end to a preset storage location;
[0007] Set up the second backend storage on the SoC side, create an operation interface, and start the NVMe subsystem;
[0008] Loading the NVMe driver on the Host side, the Host side pulls the NVMe disk through the Target side;
[0009] The test items of the NVMe disk are tested; wherein the test items include: NVMe performance, reliability, cluster configuration and data integrity.
[0010] According to another aspect of the present disclosure, a backend storage test device based on SPDK is provided, and the device is applied to a target backend storage scenario, wherein the target backend storage scenario includes: a non-volatile memory standard NVMe over Fabric scenario on a network, a small computer system interface iSCSI scenario, or a distributed file system Ceph scenario; the device includes:
[0011] A first creation module, used to create a target service and a first backend storage corresponding to the target backend storage scenario on the Target side;
[0012] A file name adding module, used for adding the file name stored in the first back-end to a preset storage location;
[0013] A second creation module is used to set a second backend storage on the system-on-chip SoC end, create an operation interface, and start the NVMe subsystem;
[0014] A disk loading module, used to load the NVMe driver on the Host side, and the Host side pulls the NVMe disk through the Target side;
[0015] A test module is used to test the test items of the NVMe disk; wherein the test items include: NVMe performance, reliability, cluster configuration and data integrity.
[0016] According to another aspect of the present disclosure, a computer-readable storage medium is provided, wherein the storage medium stores a computer program, and the computer program is used to execute the above method.
[0017] Compared with the prior art, the technical solution provided by the embodiments of the present disclosure has the following advantages:
[0018] For NVMe over Fabric scenarios, iSCSI scenarios, and Ceph scenarios, the method provided by the embodiments of the present disclosure includes: creating a target service and a first backend storage corresponding to the target backend storage scenario on the Target side; adding the file name of the first backend storage to a preset storage location; setting a second backend storage on the system-on-chip SoC side, creating an operation interface, and starting the NVMe subsystem; loading the NVMe driver on the Host side, and the Host side pulling the NVMe disk through the Target side; testing the test items of the NVMe disk; wherein the test items include: NVMe performance, reliability, cluster configuration, and data integrity.
[0019] This technical solution implements NVMe over Fabric, iSCSI and Ceph different back-end storage through SPDK, and performs various tests on high-throughput, low-latency and high-reliability data transmission. It can better implement the function and performance test of any combination of back-end storage DPU network cards, thereby improving the stability and high performance of different storage systems in actual scenarios, realizing business optimization and adjustment, and better meeting the testing needs of business scenarios. BRIEF DESCRIPTION OF THE DRAWINGS
[0020] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.
[0021] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.
[0022] Figure 1 This is a flow chart of the backend storage testing method based on SPDK described in an embodiment of the present disclosure;
[0023] Figure 2 This is an interaction diagram of the NVMe over Fabric scenario described in the embodiment of the present disclosure;
[0024] Figure 3 This is an interaction diagram in the iSCSI scenario described in the embodiment of the present disclosure;
[0025] Figure 4 This is an interaction diagram in the Ceph scenario described in the embodiment of the present disclosure;
[0026] Figure 5 This is a structural block diagram of a backend storage test device based on SPDK according to an embodiment of the present disclosure;
[0027] Figure 6 It is a schematic diagram of the structure of the electronic device described in the embodiment of the present disclosure. DETAILED DESCRIPTION
[0028] In order to more clearly understand the above-mentioned objectives, features and advantages of the present disclosure, the scheme of the present disclosure will be further described below. It should be noted that the embodiments of the present disclosure and the features in the embodiments can be combined with each other without conflict.
[0029] In the following description, many specific details are set forth to facilitate a full understanding of the present disclosure, but the present disclosure may also be implemented in other ways different from those described herein; it is obvious that the embodiments in the specification are only part of the embodiments of the present disclosure, rather than all of the embodiments.
[0030] Considering that the test method based on SPDK backend storage has the problems of complexity, platform dependence, not applicable to all application scenarios, customization requirements, and not suitable for all storage hardware, etc. The present disclosure provides a backend storage test method, device and medium based on SPDK.
[0031] To facilitate understanding, some technical contents involved in this disclosure are first introduced here.
[0032] NVMe SSD is an application software acceleration library used as backend storage, and its core is the user-mode, asynchronous, polling NVMe driver. Compared with the kernel's NVMe driver, SPDK can significantly reduce the latency of NVMe commands and improve the IOps (Input / Output Operations PerSecond, the number of read and write operations per second) of a single CPU (Central Processing Unit) core, forming a cost-effective solution. For example, SPDK's vhost (Linux / UNIX-based virtual host management system) solution can be applied to HCI (Host Controller Interface) to accelerate the NVMe I / O of virtual machines to improve the performance of storage applications. SPDK aims to solve the performance bottleneck in traditional storage systems, using user-mode drivers and modern storage technologies to provide higher performance and lower latency storage solutions. SPDK supports high-performance storage protocols such as NVMe and RDMA, and is used to build various storage applications such as distributed file systems, distributed databases, and storage acceleration caches. By moving storage management and data transfer to user space, SPDK avoids frequent context switching between kernel and user space, improves performance, and addresses the needs of large-scale data processing.
[0033] NVMe Over Fabrics uses fabric technologies such as RDMA or Fiber Channel (FC) architecture to replace PCIe transmission. Transmission based on RDMA architecture includes protocols such as Ethernet ROCE, InfiniBand, and iWARP.
[0034] The main functions of SPDK include: storage protocol layer (Storage Protocols), storage service layer (StorageServices) and driver layer (Drivers). SPDK provides a complete set of tools and libraries, which have realized high-performance, highly scalable, user-mode storage applications, aiming to significantly reduce the software overhead of the storage IO stack, thereby improving storage performance.
[0035] Based on the above embodiments, the present disclosure can realize seamless docking tests of different back-end storages in different application scenarios such as NVMe over Fabric (non-volatile memory standard on the network, NVMe-oF), iSCSI (Internet Small Computer System Interface), Ceph (distributed file system), etc., and better solve at least one of the following problems: storage performance challenges, providing higher storage performance; moving the management and data transmission of storage devices to user space to avoid switching back and forth between kernel state and user state; high-performance storage applications for building high-energy storage applications, distributed file systems, and distributed databases; building blocks and libraries: solving a series of building blocks and libraries for storage applications, NVMe, RDMA, block devices, file systems, and grouping and using them according to needs.
[0036] Figure 1 A flowchart of a backend storage testing method based on SPDK is provided in an embodiment of the present disclosure, and the method includes the following steps.
[0037] S102, creating a target service and a first backend storage corresponding to a target backend storage scenario on the Target side; the Target side is a disk array or other host equipped with a disk;
[0038] S104, adding the file name stored in the first backend to a preset storage location;
[0039] S106, setting a second backend storage on the SoC (System on Chip) side, creating an operation interface, and starting the NVMe subsystem;
[0040] S108, the NVMe driver is loaded on the Host end, and the Host end pulls the NVMe disk through the Target end; the Host end is the host end;
[0041] S110. Test the test items of the NVMe disk; wherein the test items include: NVMe performance, reliability, cluster configuration, and data integrity.
[0042] The SPDK-based backend storage testing method provided in the embodiments of the present disclosure can be executed based on a target backend storage scenario, such as a NVMe over Fabric scenario, an iSCSI scenario, and a Ceph scenario.
[0043] Next, the above SPDK-based backend storage testing method is described respectively for NVMe over Fabric scenario, iSCSI scenario and Ceph scenario.
[0044] like Figure 2 As shown, the target backend storage scenario is an NVMe over Fabric scenario; accordingly, the above S102 creates a target service and a first backend storage corresponding to the target backend storage scenario on the Target side, which may include:
[0045] Create an NVMe over Fabric service and a target file on the Target side; wherein the NVMe over Fabric service uses the TCP (Transmission Control Protocol) protocol or the RDMA (Remote Direct Memory Access) protocol for data transmission; the target file is used as the first backend storage.
[0046] Specifically, a target file is created on the Target end as the first backend storage on the Target end. An NVMe over Fabric service is built on the Target end, and the TCP protocol or the RDMA protocol is used for data transmission. Also, the file name of the first backend storage is added to the LUN (Logical Unit Number).
[0047] In this embodiment, the above step S106 sets the second backend storage on the SoC side, creates an operation interface, and starts the NVMe subsystem, which may include:
[0048] Create a large memory page on the SoC side as the second back-end storage matching the first back-end storage, configure the operation interface by loading the PCI (Peripheral Component Interconnect) bus and binding the ID (Identity document) of the PCI bus; set the PCIe (Peripheral Component Interconnect Express, a high-speed serial computer expansion bus standard) device ID and CPU core number to start the NVMe disk, and create a transmission protocol consistent with the Taget end on the SoC side, such as the NVMe over TCP protocol, to create and start the NVMe subsystem.
[0049] In a specific embodiment, a large memory page is created on the SoC side. The large memory page is allocated with physical memory and will not be swapped out to the disk by the memory management of the operating system. Therefore, there will be no page fault interruption, and it is not easy to introduce a delay in accessing the disk. The large memory page is physically continuous, and the access to the large memory page can also be accelerated.
[0050] Load the created PCI bus on the SoC side. The PCI bus has a unique ID, so bind the ID of the PCI bus to configure the operation interface.
[0051] The PCIe device ID and the number of CPU cores are bound one by one. In this case, this embodiment sets the PCIe device ID and the number of CPU cores bound thereto, such as 8 cores, 16 cores, etc. Different numbers of CPU cores correspond to different performances; after binding, the NVMe system is started.
[0052] Then, a transmission protocol NVMe over TCP is created on the SoC side to transmit data on the SoC side. For example, data on a remote NVMe disk can be accessed through the NVMe over TCP protocol.
[0053] Based on the started NVMe system, the SoC first uses the PF (Physical Function) interface to create an NVMe disk corresponding to the back-end storage of the NVMe subsystem in the NVMe system, and then creates a virtual VF (Virtual Function) interface corresponding to the PF interface, and uses the VF interface to create the NVMe disk of the back-end storage.
[0054] On the SoC side, after configuring the above large memory pages, operation interfaces, NVMe disks, and transmission protocols, the creation of the NVMe subsystem is completed.
[0055] In this embodiment, the NVMe driver is loaded on the Host side, and the Host side pulls the NVMe disk through the Target side; the test items of the NVMe disk are tested through a preset test tool; wherein the test tool is, for example: FIO and IOMeter, etc., FIO is Flexible I / O Tester, which is a very good tool for testing server IOPS, and can be used for stress testing and verification of hard disks; IOMeter is a tool for testing disk performance, which can test the I / O transmission rate and average I / O response time.
[0056] At this point, the test in the NVMe over Fabric backend storage scenario is implemented. In this embodiment, the scalability of the NVMe overFabric backend storage scenario can include: (1) flexible expansion of storage nodes, easy expansion of storage capacity and performance according to customer needs; (2) storage nodes can be NVMe SSDs, NVMe-oF storage controllers, and storage devices of other protocols; Host flexible expansion, flexible addition of more hosts and servers to expand the computing and storage capabilities of the NVMe-oF system; Host can flexibly access storage nodes through network connections and use the NVMe protocol to access data; (3) support for multi-path access to prevent downtime from causing inaccessibility and load balancing. NVMe-oF has excellent scalability and can adapt to growing storage needs.
[0057] like Figure 3 As shown, the target backend storage scenario is an iSCSI scenario; accordingly, the above S102 creates a target service and a first backend storage corresponding to the target backend storage scenario on the Target side, which may include: creating an iSCSI service and a disk fdisk on the Target side; wherein the disk fdisk serves as the first backend storage.
[0058] Then, you can perform iSCSI authentication on the disk fdisk. After the authentication is passed, view the created disk fdisk and add the file name of the disk fdisk to the iSCSI LUN.
[0059] For step S106, a second backend storage is set on the SoC side and an operation interface is created. This embodiment may include: loading SPDK iSCSI Bdev on the SoC side as the second backend storage matching the first backend storage, and creating iSCSI BdeviSCSI0 as an operation interface; wherein SPDK iSCSI Bdev is a general block layer after abstraction of devices such as NVMe in SPDK.
[0060] After creating the iSCSI service and the first backend storage on the Target side, and setting the second backend storage and operation interface on the SoC side, start the NVMe subsystem. Load the NVMe driver on the Host side, and the Host side pulls the NVMe disk through the Target side; test the test items of the NVMe disk through the preset test tool.
[0061] At this point, the test in the iSCSI backend storage scenario is implemented. In this embodiment, the scalability of the iSCSI backend storage scenario may include: (1) multipath and redundancy, supporting the configuration expansion of multipath to increase the redundancy and availability of network links; (2) supporting the expansion of iSCSI Initiator and Target load balancing, supporting load balancing in large-scale iSCSI deployment to ensure load balancing between each iSCSI Initiator and Target; and supporting multiple Initiators to allocate multiple Targets; (3) storage capacity expansion, the storage space can flexibly expand capacity according to demand, add more iSCSI Target devices or expand the storage pool to increase storage capacity; it can support virtualization technology (LUN) to more flexibly achieve capacity expansion and more flexibly manage storage resources.
[0062] like Figure 4 As shown, the target backend storage scenario is a Ceph scenario; accordingly, the above S102 creates a target service and a first backend storage corresponding to the target backend storage scenario on the Target side, which may include: creating a Ceph service on the Target side, entering the cephadm console to add multiple disks; wherein the multiple disks serve as the first backend storage.
[0063] Furthermore, this embodiment creates a storage pool for recording multiple disk addresses in the cephadm console.
[0064] In a specific embodiment, the Target side builds a Ceph service; the Target side enters the cephadm console to add multiple disks as the first backend storage, and the above disks can be RBD (RADOS Block Device). In this embodiment, the host in the cluster can also be checked through the cephadm console, and the Ceph service and the first backend storage can be checked through the host to see whether they are successfully created.
[0065] Create a storage pool in the cephadm console. The storage pool is used to record the addresses of multiple RBD disks. In other words, the storage pool is the partition storage space of multiple RBD disks. Add the names of multiple disks to the storage pool.
[0066] In this embodiment, a large memory page is created on the SoC side as the second backend storage matching the first backend storage, and an RBD Bdev is created as an operation interface. Also, the NVMe driver is loaded on the Host side, and the Host side pulls the NVMe disk through the Target side; and the test items of the NVMe disk are tested through a preset test tool.
[0067] At this point, the test in the Ceph backend storage scenario is implemented. In this embodiment, the scalability of the Ceph backend storage scenario may include: (1) a flexible distributed architecture, Ceph data storage and management are distributed on multiple nodes (OSD), monitoring nodes (MON) and metadata servers (MDS); storage capacity and performance can be expanded by adding new nodes, so that existing services should not be interrupted; (2) there are flexible data sharding and replication operations, Ceph shards and replicates data to multiple storage nodes to improve performance and data redundancy; more storage nodes can be added simply and flexibly to increase the amount of data sharding and replication to meet the needs of the scenario.
[0068] Based on the above embodiments, the test items of the NVMe disk can be tested as shown below.
[0069] Regarding flexible performance testing. This embodiment can measure the throughput, IOPS and latency of NVMe-oF storage through a preset test tool to evaluate the load and data transmission mode of the back-end storage in the target back-end storage scenario, and represent the performance of NVMe-oF according to the evaluation results of the load and data transmission mode.
[0070] Regarding reliability testing, Yusur DPU hardware or network failures were used to evaluate the fault tolerance and recovery capabilities of the NVMe-oF system, and the reconstruction time and data integrity protection functions were achieved.
[0071] Cluster configurability testing: Through the configurability of the Ceph cluster, node settings are implemented to ensure that performance and reliability requirements are met.
[0072] Tests on data integrity: The scalability protection function of the Ceph cluster is reliable, data verification and fault tolerance are achieved, DPU functions can be better realized, higher performance and minimum latency can be achieved, and data reading and writing of NVMe disks can be realized.
[0073] In summary, for NVMe over Fabric scenarios, iSCSI scenarios, and Ceph scenarios, the method provided by the embodiments of the present disclosure includes: creating a target service and a first backend storage corresponding to the target backend storage scenario on the Target side; adding the file name of the first backend storage to a preset storage location; setting a second backend storage on the SoC side, creating an operation interface, and starting the NVMe subsystem; loading the NVMe driver on the Host side, and the Host side pulls the NVMe disk through the Target side; testing the test items of the NVMe disk; wherein the test items include: NVMe performance, reliability, cluster configuration, and data integrity.
[0074] This technical solution implements NVMe over Fabric, iSCSI and Ceph different back-end storage through SPDK, and performs various tests on high-throughput, low-latency and high-reliability data transmission. It can better implement the function and performance test of any combination of back-end storage DPU network cards, thereby improving the stability and high performance of different storage systems in actual scenarios, realizing business optimization and adjustment, and better meeting the testing needs of business scenarios.
[0075] Figure 5 A structural block diagram of a backend storage testing device based on SPDK provided in an embodiment of the present disclosure, the device includes the following modules.
[0076] A first creation module 210, configured to create a target service and a first backend storage corresponding to the target backend storage scenario on the Target side;
[0077] A file name adding module 220, used to add the file name stored in the first back-end to a preset storage location;
[0078] The second creation module 230 is used to set the second backend storage on the SoC side, create an operation interface, and start the NVMe subsystem;
[0079] The disk loading module 240 is used to load the NVMe driver on the Host side, and the Host side pulls the NVMe disk through the Target side;
[0080] The test module 250 is used to test the test items of the NVMe disk; wherein the test items include: NVMe performance, reliability, cluster configuration and data integrity.
[0081] In some embodiments, the device is applied to a target backend storage scenario, and the target backend storage scenario includes: NVMe over Fabric scenario, iSCSI scenario and Ceph scenario.
[0082] In some embodiments, the target backend storage scenario is the NVMe over Fabric scenario; the first creation module 210 is further used to:
[0083] Create an NVMe over Fabric service and a target file on the Target end; wherein the NVMe over Fabric service uses the TCP protocol or the RDMA protocol for data transmission; and the target file is used as the first backend storage;
[0084] The second creation module 230 is further used for:
[0085] Creating a large page memory on the SoC side as the second backend storage matching the first backend storage, and configuring an operation interface by loading a PCI bus and binding an ID of the PCI bus;
[0086] The device ID and the number of CPU cores of the high-speed serial computer expansion bus standard PCIe are set to start the NVMe system, and a transmission protocol consistent with the Taget end is created on the SoC end to create and start the NVMe subsystem.
[0087] In some embodiments, after starting the NVMe system, the device further includes:
[0088] Using the PF interface on the SoC side, creating an NVMe disk corresponding to the backend storage of the NVMe subsystem in the NVMe system;
[0089] Create a virtual VF interface corresponding to the PF interface, and use the VF interface to create an NVMe disk for back-end storage.
[0090] In some embodiments, the target backend storage scenario is the iSCSI scenario;
[0091] The first creation module 210 is also used to: create an iSCSI service and a disk fdisk on the Target side; wherein the disk fdisk serves as the first backend storage;
[0092] The file name adding module 220 is also used for: performing iSCSI authentication on the disk fdisk, and after the authentication is passed, adding the file name of the disk fdisk to the iSCSI LUN;
[0093] The second creation module 230 is further used for: loading SPDK iSCSI Bdev on the SoC side as the second backend storage matching the first backend storage, and creating iSCSI Bdev iSCSI0 as an operation interface.
[0094] In some embodiments, the target backend storage scenario is the Ceph scenario;
[0095] The first creation module 210 is also used to: create a Ceph service on the Target end, enter the cephadm console to add multiple disks; wherein the multiple disks serve as the first backend storage;
[0096] The file name adding module 220 is also used for: creating a storage pool recording the addresses of the multiple disks in the cephadm console;
[0097] The second creation module 230 is further used to: create a large memory page on the SoC side as the second backend storage matching the first backend storage, and create an RBD Bdev as an operation interface.
[0098] In some embodiments, the device further includes: checking the hosts in the cluster through the cephadm console, and checking through the hosts whether the Ceph service and the first backend storage are created successfully.
[0099] In some embodiments, the testing module 250 is further configured to:
[0100] The throughput, IOPS and latency of the NMVe-oF storage are measured by a preset test tool to evaluate the load and data transmission mode of the back-end storage in the target back-end storage scenario, and the performance of the NVMe-oF is represented according to the evaluation results of the load and data transmission mode.
[0101] The implementation principle and technical effects of the device provided in this embodiment are the same as those of the aforementioned method embodiment. For the sake of brief description, for matters not mentioned in the device embodiment, reference may be made to the corresponding contents in the aforementioned method embodiment.
[0102] Figure 6 The structure diagram of an electronic device provided by the embodiment of the present disclosure is shown in FIG. Figure 6 As shown, the electronic device 300 includes one or more processors 301 and a memory 302 .
[0103] The processor 301 may be a central processing unit (CPU) or other forms of processing units having data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device 300 to perform desired functions.
[0104] The memory 302 may include one or more computer program products, and the computer program product may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory (cache), etc. The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 301 may run the program instructions to implement the SPDK-based backend storage test method of the embodiment of the present disclosure described above and / or other desired functions. Various contents such as input signals, signal components, noise components, etc. may also be stored in the computer-readable storage medium.
[0105] In one example, the electronic device 300 may further include: an input device 303 and an output device 304 , and these components are interconnected via a bus system and / or other forms of connection mechanisms (not shown).
[0106] In addition, the input device 303 may also include, for example, a keyboard, a mouse, and the like.
[0107] The output device 304 can output various information to the outside, including the determined distance information, direction information, etc. The output device 304 can include, for example, a display, a speaker, a printer, a communication network and a remote output device connected thereto, and the like.
[0108] Of course, to simplify, Figure 6 Only some of the components related to the present disclosure in the electronic device 300 are shown, and components such as a bus, an input / output interface, etc. are omitted. In addition, according to specific application situations, the electronic device 300 may also include any other appropriate components.
[0109] Furthermore, this embodiment also provides a computer-readable storage medium, wherein the storage medium stores a computer program, and the computer program is used to execute the above-mentioned SPDK-based backend storage testing method.
[0110] The embodiments of the present disclosure provide a computer program product of a backend storage testing method, device, electronic device, and medium based on SPDK, including a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the methods described in the previous method embodiments. For specific implementation, please refer to the method embodiments, which will not be repeated here.
[0111] It should be noted that, in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.
[0112] The above description is only a specific embodiment of the present disclosure, so that those skilled in the art can understand or implement the present disclosure. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present disclosure. Therefore, the present disclosure will not be limited to the embodiments described herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A backend storage testing method based on the storage performance development toolkit SPDK, characterized in that: The method comprises: Create a target service and a first backend storage corresponding to the target backend storage scenario on the Target side of the disk array; Adding the file name stored in the first back-end to a preset storage location; Setting up the second back-end storage on the system-on-chip SoC, creating an operation interface, and starting the non-volatile memory host controller interface specification NVMe subsystem; Loading the NVMe driver on the host Host, the Host pulls the NVMe disk through the Target; Testing the test items of the NVMe disk; wherein the test items include: NVMe performance, reliability, cluster configuration and data integrity; The target backend storage scenario is an NVMe over Fabric scenario; The step of creating a target service and a first backend storage corresponding to the target backend storage scenario on the Target side includes: Create an NVMe over Fabric service and a target file on the Target end; wherein the NVMe over Fabric service uses the Transmission Control Protocol TCP protocol or the Remote Direct Data Access RDMA protocol for data transmission; and the target file is used as the first backend storage; The step of setting the second backend storage on the SoC side, creating an operation interface, and starting the NVMe subsystem includes: Create a large page memory on the SoC side as the second backend storage matching the first backend storage, and configure the operation interface by loading the peripheral device interconnection PCI bus and binding the identity ID of the PCI bus; The device ID and the number of CPU cores of the high-speed serial computer expansion bus standard PCIe are set to start the NVMe system, and a transmission protocol consistent with the Target end is created on the SoC end to create and start the NVMe subsystem.
2. The method according to claim 1, characterized in that The method is applied to a target backend storage scenario, wherein the target backend storage scenario includes: a non-volatile memory standard NVMe over Fabric scenario on a network, a small computer system interface iSCSI scenario, and a distributed file system Ceph scenario.
3. The method according to claim 1, characterized in that After starting the NVMe system, the method further includes: Using a physical device function PF interface on the SoC side to create an NVMe disk corresponding to the backend storage of the NVMe subsystem in the NVMe system; A virtual virtual device function VF interface corresponding to the PF interface is created, and the VF interface is used to create an NVMe disk for back-end storage.
4. The method according to claim 1, characterized in that: The target backend storage scenario is an iSCSI scenario; The step of creating a target service and a first backend storage corresponding to the target backend storage scenario on the Target side includes: Create an iSCSI service and a disk fdisk on the Target side; wherein the disk fdisk serves as the first backend storage; The step of adding the file name stored in the first backend to a preset storage location includes: Performing iSCSI authentication on the disk fdisk, and after passing the authentication, adding the file name of the disk fdisk to the iSCSI logical unit number; The step of setting the second backend storage on the SoC side and creating an operation interface includes: On the SoC side, SPDK iSCSI Bdev is loaded as the second backend storage matching the first backend storage, and iSCSI Bdev iSCSI0 is created as the operation interface.
5. The method according to claim 1, characterized in that The target backend storage scenario is a Ceph scenario; The step of creating a target service and a first backend storage corresponding to the target backend storage scenario on the Target side includes: Create a Ceph service on the Target side, enter the cephadm console and add multiple disks; wherein the multiple disks serve as the first backend storage; Adding the file name of the first backend storage to a preset storage location includes: Create a storage pool recording the addresses of the multiple disks in the cephadm console; The step of setting the second backend storage on the SoC side and creating an operation interface includes: A large memory page is created on the SoC side as the second backend storage matching the first backend storage, and RBDBdev is created as an operation interface.
6. The method according to claim 5, characterized in that The method further comprises: Check the hosts in the cluster through the cephadm console, and check through the hosts whether the Ceph service and the first backend storage are created successfully.
7. The method according to claim 1, characterized in that The testing of the test items of the NVMe disk includes: The throughput, number of read and write operations per second (IOPS) and latency of the NMVe-oF storage are measured by a preset test tool to evaluate the load and data transmission mode of the back-end storage in the target back-end storage scenario, and the performance of the NVMe-oF is represented based on the evaluation results of the load and data transmission mode.
8. A backend storage testing device based on the storage performance development toolkit SPDK, characterized in that: The device comprises: A first creation module is used to create a target service and a first backend storage corresponding to a target backend storage scenario on a disk array Target end; A file name adding module, used for adding the file name stored in the first back-end to a preset storage location; A second creation module is used to set a second back-end storage on the system-on-chip SoC end, create an operation interface, and start a non-volatile memory host controller interface specification NVMe subsystem; A disk loading module, used to load the NVMe driver on the host Host end, and the Host end pulls the NVMe disk through the Target end; A test module, used to test the test items of the NVMe disk; wherein the test items include: NVMe performance, reliability, cluster configuration and data integrity; The target backend storage scenario is an NVMe over Fabric scenario; The first creation module is also used for: Create an NVMe over Fabric service and a target file on the Target end; wherein the NVMe over Fabric service uses the Transmission Control Protocol TCP protocol or the Remote Direct Data Access RDMA protocol for data transmission; and the target file is used as the first backend storage; The second creation module is also used for: Create a large page memory on the SoC side as the second backend storage matching the first backend storage, and configure the operation interface by loading the peripheral device interconnection PCI bus and binding the identity ID of the PCI bus; The device ID and the number of CPU cores of the high-speed serial computer expansion bus standard PCIe are set to start the NVMe system, and a transmission protocol consistent with the Target end is created on the SoC end to create and start the NVMe subsystem.
9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores instructions, and when the instructions are executed on a terminal device, the terminal device implements the method according to any one of claims 1 to 7.
