A virtualized storage pool based on heterogeneous instruction sets and method thereof

Through a virtualized storage pool based on heterogeneous instruction sets, the Ceph protocol and intermediate layer SG realize redundancy and dynamic migration of storage nodes in different instruction sets, the path dependence problem of a single instruction set storage resource pool is solved, and the utilization of storage resources is improved The rate and system's ability to resist disfigurement ensure the independent controllability of domestic products.

CN115065693BActive Publication Date: 2024-11-22SICHUAN INDEPENDENT CONTROLLABLE ELECTRONIC INFORMATION IND CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210661378.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-06-13
Publication Date
2024-11-22
Estimated Expiration
2042-06-13

AI Technical Summary

Technical Problem

The storage resource pool in the prior art only supports a single instruction set, which causes the service system to stagnate, downgrade or interrupt when a "bottleneck" event occurs on a server or storage device of a certain technical route, and the CPU chip is out of supply. affects the normal operation of the business.

Method used

Using a virtualized storage pool based on heterogeneous instruction set, it provides block storage, object storage and file storage protocols through the protocol layer, local API processing layer and heterogeneous distributed storage core layer, and uses Ceph protocol and intermediate layer SG to implement different instructions. Communication and redundancy between storage data nodes of the set, dynamically migrate storage data nodes to improve the reliability and scalability of the system.

Benefits of technology

It reduces the path dependence on a single instruction set, improves the utilization rate of storage resource pools, realizes support for diversified computing needs, and ensures the storage system's anti-disaster disaster ability and domestic independent controllability.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115065693B_ABST
    Figure CN115065693B_ABST
Patent Text Reader

Abstract

The present invention discloses a virtualized storage pool based on a heterogeneous instruction set and a method thereof, belonging to the technical field of information technology application innovation. The virtualized storage pool comprises: a protocol layer, the protocol layer is used to provide block storage, object storage and file storage protocols, and data read and write commands are input through the block storage, object storage and file storage protocols; a local API processing layer, the local API processing layer is used to provide library functions, and data read and write commands are executed by calling the library functions; a heterogeneous distributed storage core layer, the heterogeneous distributed storage core layer provides storage data nodes of different instruction sets, and each storage data node establishes communication with the local API processing layer through the Ceph protocol, and at least three storage data nodes are combined into a server, and each different server corresponds to storage read and write data, so as to achieve the purpose of not only reducing the path dependence on a single instruction set, but also improving the utilization rate of the storage resource pool.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of information technology application innovation, and in particular, relates to a virtualized storage pool based on heterogeneous instruction sets and a method thereof. Background Art

[0002] At present, information technology application innovation (abbreviated as: Xinchuang, the same below) has become a national strategy and a new driving force for national economic development under the current situation. The development of Xinchuang is to solve the problem of intrinsic safety. Intrinsic safety means that we should first make it something that we can control, study, develop and produce.

[0003] With the development of the information and innovation industry, as an important direction of new infrastructure, domestic CPUs and operating systems are the core of achieving independent control and the key link to solving the "bottleneck". The main participants in the domestic CPU are brands such as Loongson, Zhaoxin, Feiteng, Hygon, Shenwei and Huawei Kunpeng. A technical route based on four major mainstream instruction sets such as LoongArch (Loongson), alpha (Shenwei), ARM (Feitian, Kunpeng), and x86 (Haiguang, Zhaoxin) has gradually been formed, becoming the key to the digital transformation of the economy and the promotion of the development of the industrial chain. Starting from the introduction of technology systems, strengthening the industrial foundation, and strengthening the guarantee capabilities, it promotes the localization of the information and innovation industry, drives the transformation of the traditional IT information industry, and builds a regional industrial cluster.

[0004] Based on this, the current construction of party and government or industry data centers (IDC), whether public or private, will fully consider the risks of policies, application environment, personnel, etc., and formulate corresponding risk countermeasures and management measures, and formulate sustainable design plans for the situation where different domestic chip architectures exist in the basic environment. The infrastructure covered includes domestic chip servers with four major instruction sets, domestic storage devices, domestic network devices and domestic security devices. Based on such hardware implementation, the corresponding underlying support is configured, including operating systems, virtualization software, etc., to form computing resource pools (virtual machine pools, bare metal physical machine pools), storage resource pools (FC-SAN storage resource pools, distributed storage resource pools), network resource pools (VPN, elastic IP, virtual switches), etc., to build an infrastructure as a service layer (IAAS).

[0005] However, the industry's current IAAS-level storage resource pools only support a single instruction set technology route, and there is a certain path dependency. When a server, storage or network switch, or router of a certain technology route encounters a "bottleneck" event such as suspension of supply, suspension of operation and maintenance, or suspension of upgrades, the business system will stagnate, degrade, or even be interrupted.

[0006] Although there are some storage virtualization technologies based on large-scale distributed parallel I / O of a single domestic chip, once a "supply cut" occurs at the CPU chip level, it will greatly affect the normal operation of the business. Summary of the invention

[0007] In view of this, in order to solve the above-mentioned problems existing in the prior art, the purpose of the present invention is to provide a virtualized storage pool and method based on heterogeneous instruction sets so as to not only reduce the path dependence on a single instruction set but also improve the utilization rate of the storage resource pool.

[0008] The technical solution adopted by the present invention is: a virtualized storage pool based on a heterogeneous instruction set, the virtualized storage pool comprising:

[0009] The protocol layer is used to provide block storage, object storage and file storage protocols, and input data read and write commands through the block storage, object storage and file storage protocols;

[0010] A local API processing layer, wherein the local API processing layer is used to provide library functions and execute data read and write commands by calling library functions;

[0011] A heterogeneous distributed storage core layer provides storage data nodes with different instruction sets, and each storage data node establishes communication with the local API processing layer through the Ceph protocol, and is combined into a server by at least three storage data nodes, and each different server corresponds to storage read and write data.

[0012] Further, the block storage communication is connected with a distributed system interface; the object storage communication is connected with a bare metal server interface and / or a virtual machine interface; and the file storage protocol communication is connected with a POSIX file system interface.

[0013] Furthermore, the heterogeneous distributed storage core layer includes:

[0014] A client, wherein the client establishes a communication connection with a library function of a local API processing layer;

[0015] An object storage device, wherein the object storage device comprises a plurality of storage data nodes, each of the storage data nodes is deployed on a server with a different instruction set, and a communication connection is established between each of the storage data nodes;

[0016] A monitoring and maintenance device is used to establish a communication connection with each of the storage data nodes and exchange status information of each of the storage data nodes.

[0017] Furthermore, the servers of different instruction sets are: LoongArch server, alpha server, ARM server and x86 server.

[0018] The present invention also provides a virtualized storage method based on a heterogeneous instruction set, the storage method comprising:

[0019] S1: Call library functions to execute read and write commands through the user end, block storage, object storage, or file storage;

[0020] S2: If the library function reads a file through the client, it enters the file reading process; if the library function writes a file through the client, it enters the file writing process;

[0021] The write file includes:

[0022] A1: The client splits the file into several object storage blocks of the same size.

[0023] A2: Map each object storage block to n mutually redundant storage data nodes, n ≥ 2;

[0024] A3: Initiate a write request to the first-level storage data node among the n storage data nodes;

[0025] A4: Synchronously replicate n-1 copies to n-1 storage data nodes with different instruction sets;

[0026] A5: Confirm the writing of the object storage block to the client;

[0027] The reading file includes:

[0028] B1: Get the index node from the file and calculate the storage data node where the index node is located based on the index node;

[0029] B2: initiates a read request to the storage data node;

[0030] B3: Returns the read object storage block to the client.

[0031] Furthermore, the A1 includes:

[0032] A101: split the file into several object storage blocks;

[0033] A102: Number several object storage blocks to generate heterogeneous OIDs.

[0034] Furthermore, the above A2 includes:

[0035] A201: Map each object storage block to an intermediate layer SG, and organize the object storage blocks through the intermediate layer SG, so that a "one-to-many" mapping relationship is formed between the intermediate layer SG and the object storage blocks;

[0036] A202: Mapping an intermediate layer SG to a plurality of storage data nodes by mapping the intermediate layer SG to the storage data nodes, so that the intermediate layer SG can be dynamically migrated to a combination of different storage data nodes.

[0037] Further, in A202, when the intermediate layer SG is allocated to multiple storage data nodes, each storage data node is a server of a different instruction set, or a server of the same instruction set.

[0038] Furthermore, B2 is: initiating a read request to a storage data node, and if the storage data node is damaged, sending a read request to other storage data nodes that are redundant to the storage data node according to priority until the object storage block is read.

[0039] The beneficial effects of the present invention are:

[0040] 1. The virtualized storage pool and method based on heterogeneous instruction sets provided by the present invention are adopted. By adopting a virtualized storage pool based on heterogeneous instruction sets such as LoongArch, alpha, ARM, x86, etc., the underlying distributed storage processing mechanism of the entire data center is completed. Distributed storage means that it can be deployed on multiple servers and provide high-performance reading and writing to the outside world through parallel processing of multiple servers. This technology provides technology that supports heterogeneous instruction sets, solves the dependence of traditional Ceph protocols on a single technical route, and solves the problem of being "stuck" by others in virtualized storage pools in the past.

[0041] 2. The virtualized storage pool based on heterogeneous instruction sets and the method thereof provided by the present invention can be widely used in cloud computing IDC, big data storage base, data management, smart cities and other applications with large data volumes and requiring domestically produced independent and controllable applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is an overall architecture diagram of a virtualized storage pool based on heterogeneous instruction sets provided by the present invention;

[0043] Figure 2 It is a schematic diagram of mapping of a heterogeneous distributed storage core layer in a virtualized storage pool based on heterogeneous instruction sets provided by the present invention;

[0044] Figure 3 It is a schematic diagram of the workflow of the virtualized storage method based on heterogeneous instruction sets provided by the present invention. DETAILED DESCRIPTION

[0045] Embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar modules or modules with the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be construed as limitations on the present application. On the contrary, the embodiments of the present application include all changes, modifications and equivalents that fall within the spirit and connotation of the appended claims.

[0046] Example 1

[0047] In this embodiment, a virtualized storage pool based on heterogeneous instruction sets is specifically provided. The virtualized storage pool is based on the virtualized storage pool technology of heterogeneous instruction sets, which distributes files, objects and data blocks to the disks of servers or workstations with different instruction sets through distributed storage technology. Specifically, it is based on the virtualized storage pool technology of heterogeneous instruction sets such as LoongArch, alpha, ARM, and x86, and provides unified storage services to the outside world.

[0048] like Figure 1 As shown, the virtualized storage pool includes a combination of three parts: a protocol layer, a local API processing layer, and a heterogeneous distributed storage core layer. Among them, the protocol layer is used to provide block storage, object storage, and file storage protocols, and input data read and write commands through block storage, object storage, and file storage protocols to shield the differences of the underlying CPU servers with different instruction sets. In actual application, the block storage communication is connected with a distributed system interface, such as: Amazon S3 distributed system and Swift distributed system; the object storage communication is connected with a bare metal server (Host) interface and / or a virtual machine (VM) interface; the file storage protocol communication is connected with a POSIX file system interface, such as: NFS file system interface, CIFS file system interface, etc.

[0049] The local API processing layer is used to provide library functions and execute data read and write commands by calling library functions. This layer provides a bunch of library functions for development. These library functions and applications are generally on the same node, so it is also called the local API processing layer.

[0050] The heterogeneous distributed storage core layer provides storage data nodes with different instruction sets, and each storage data node establishes communication with the local API processing layer through the Ceph protocol, and is composed of at least three storage data nodes into a server, and each different server corresponds to storage read and write data. The heterogeneous distributed storage core layer is the core of this virtual storage pool and the place where the bottom-level data is actually stored. It is physically composed of a large number of servers with different instruction sets, and can follow the Ceph protocol to achieve reliability, automation, scalability and other characteristics.

[0051] In view of the above, in order to further illustrate the core of the present technology, the heterogeneous distributed storage core layer includes: a client, an object storage device and a monitoring and maintenance device.

[0052] The client is also called Client, which can be a program or a command line. The client establishes a communication connection with the library function of the local API processing layer, so the upper library function must interact with the storage node through the client program.

[0053] Object storage devices are also called OSDs. The object storage devices include multiple storage data nodes, which are also called OSDs. Each storage data node is deployed on a server with a different instruction set, and a communication connection is established between each storage data node. According to the Ceph protocol, each storage data node (OSD) is actually a server with an operating system and a file system installed. Generally speaking, an OSD includes at least: a single-core CPU, memory, a hard disk, and a network card. Several OSDs are deployed on servers with CPUs of different instruction sets. The servers of different instruction sets are: LoongArch server, alpha server, ARM server, and x86 server. At the same time, each OSD has a daemon thread (deamon), which is used to introduce the access connection of the client (Client), communicate with the monitoring and maintenance device (monitor) and other OSDs, and perform data storage and maintenance with other OSDs. In other words, the logical function of the OSD is completed by the daemon thread (deamon).

[0054] The monitoring and maintenance equipment is also called Monitors, which is mainly used for system status detection and maintenance. The monitoring and maintenance equipment establishes a communication connection with each of the storage data nodes and exchanges the status information of each of the storage data nodes. Each storage data node (OSD) will exchange node status information with the monitoring and maintenance equipment (monitor) to form global metadata, namely Clustermap, which can be used to obtain the specific location of data storage.

[0055] In actual applications, if a client (Client) comes with a request, no matter whether the request is for reading or writing, it needs to be addressed first to find where the data should be stored or where it needs to be read from. According to the Ceph protocol, the addressing method is calculated based on metadata (Clustermap) to avoid table lookups and the use of a separate metadata server.

[0056] In this embodiment, the use of heterogeneous instruction sets facilitates the expansion of the storage pool and can support various complex technical routes, meeting diverse computing needs, while also achieving fairness in the overall allocation of storage resources.

[0057] Since several storage data nodes (OSD) are deployed on servers with different instruction set CPUs, it also supports storage device management of servers with different instruction sets, including registration, allocation and recycling, to form a unified resource pool with the following features: First, it supports isolation of files of various business systems, and a large number of file systems can exist at the same time; second, file copies and file shards are automatically distributed throughout the system, without being limited by the instruction set and storage capacity of a single server CPU chip; third, it is globally consistent and transparent, regardless of the instruction set of LoongArch server, alpha server, ARM server or x86 server, the file system accessed from any server is consistent; fourth, each server is a file read and write interface, supporting large-scale distributed parallel I / O to ensure that the storage system itself is resistant to damage and disasters.

[0058] Example 2

[0059] On the basis of the virtualized storage pool based on heterogeneous instruction sets provided in Example 1, this embodiment further provides a virtualized storage method based on heterogeneous instruction sets, and the storage method includes:

[0060] S1: Call library functions to execute read and write commands through the user end, block storage, object storage, or file storage;

[0061] S2: If the library function writes a file through the client, it enters the file writing process; assuming a file (File) is sent from the client (Client), the file writing process includes:

[0062] A1: According to the idea of ​​distributed object storage, the client divides the file into several object storage blocks of the same size, and maps these object storage blocks to the storage data node (OSD); specifically:

[0063] A101: Split the file into several object storage blocks, that is, split the file according to a certain size, which is equivalent to the striping process in RAID;

[0064] A102: After the splitting is completed, several object storage blocks are numbered to generate heterogeneous oids (Object IDs).

[0065] According to the above, each file has a unique index node number ino, also known as metadata, and then the serial numbers generated after the file is split are concatenated together to form a heterogeneous oid. For example, a file with an index node number of 99999 is split into three object storage blocks, and the heterogeneous oids are 999990, 999991, and 999992.

[0066] Different from the traditional fixed mapping algorithm, the fixed mapping algorithm is: an object storage block (Object) is fixedly mapped to a group of storage data nodes (OSD) each time. At the same time, in order to ensure redundancy, the same object storage block (Object) will be mapped to 3 to 5 storage data nodes (OSD) and users can set different policies. Therefore, the current storage data node (OSD) will regularly interact with other storage data nodes (OSD). The object storage blocks (Object) carried on each storage data node (OSD) may reach millions. If information interaction is to be performed on storage data nodes (OSD) with different instruction sets, the amount of data will surge to tens of millions of times, and the maintenance cost will be very high.

[0067] In order to achieve dynamic mapping between object storage blocks (Object) and storage data nodes (OSD) of different instruction sets, and to reduce the information maintenance between storage data nodes (OSD) and object storage blocks (Object), an intermediate layer SG (Strategy Group) is introduced. The intermediate layer SG will first organize the objects as follows:

[0068] A2: Map each object storage block to n mutually redundant storage data nodes, where n≥2. In this embodiment, n=3.

[0069] A201: Map each object storage block to an intermediate layer SG, and organize the object storage blocks through the intermediate layer SG, forming a "one-to-many" mapping relationship between the intermediate layer SG and the object storage block; specifically: an object storage block (Object) can only be mapped to one intermediate layer SG, and an intermediate layer SG can organize multiple object storage blocks (Object), that is, there is a "one-to-many" mapping relationship between the intermediate layer SG and the object storage block (Object). In order to achieve an approximately uniform total amount of data stored in each intermediate layer SG, the details are as follows:

[0070] Each object storage block (Object) must be independently mapped to an intermediate layer SG, so that the object storage blocks (Object) can be evenly distributed in the storage data nodes (OSD) of different instruction sets, which can fully utilize the underlying hardware and spread the risk. The implementation process is as follows: First, the oid is hashed to obtain a pseudo-random value with an approximate uniform distribution, and then an intermediate layer SG is selected according to the algorithm and the object storage block (Object) is placed in it. The specific algorithm is as follows: Assuming that the number of intermediate layer SGs is m, then a mask with a length of m-1 is set, and then the pseudo-random value obtained by the hash value is bitwise ANDed with the mask, and finally the SG sequence number (sgid) is obtained, which ensures the approximately uniform mapping between the object storage block (Object) and the intermediate layer SG. At the same time, the size of the object storage block (Object) is the same, ensuring that the total amount of data stored in the object storage block (Object) in the intermediate layer SG is approximately uniform.

[0071] A202: By mapping the middle-layer SG and the storage data node (OSD), one middle-layer SG is mapped to multiple storage data nodes (OSD) so that the middle-layer SG can be dynamically migrated to a combination of different storage data nodes. For example: each middle-layer SG will be replicated into 3 copies and placed on 3 storage data nodes (OSD), and each storage data node (OSD) will also carry multiple middle-layer SGs. The details are as follows:

[0072] The mapping between the storage object storage block (Object) and the middle layer SG is static, that is, a storage object storage block (Object) will be mapped to a certain middle layer SG through this operation. The purpose of adding the middle layer SG is to realize the dynamic migration of storage data nodes (OSD) with different instruction sets. Therefore, the mapping of the middle layer SG to the storage data node (OSD) is affected by other factors. When the status and number of storage data nodes (OSD) in the system change, the mapping relationship will also change. Through a metadata (Cluster Map), the current system status, that is, the changes that may occur at present, is recorded. At the beginning, the administrator can configure some policies, for example, a middle layer SG can be allocated to 3 or 5 storage data nodes (OSD). These 3 or 5 storage data nodes (OSD) can be servers with different instruction sets, or servers or racks with the same instruction set. Therefore, when a server fails, there are other copies available.

[0073] A3: Initiate a write request to the primary storage data node among the three storage data nodes. Regardless of block storage, file storage or object storage, when a virtual machine, physical machine or user directly calls a library function to execute a read or write command, the local API processing layer will read or write a file (File) to the heterogeneous distributed storage core layer through the client (Client). The client (Client) must map the file (File) to several object storage blocks (Object) locally, and then use the Hash and Crush algorithm to map them to three storage data nodes (OSD). The storage data node (OSD) with the earliest sequence number is the primary storage data node (Primary OSD), and the next two are the secondary storage data node (Secondary OSD) and the tertiary storage data node (Tertiary OSD). The primary storage data node (Primary OSD), the secondary storage data node (Secondary OSD) and the tertiary storage data node (Tertiary OSD) are called mutually redundant.

[0074] A4: Synchronously replicate two copies to two other storage data nodes with different instruction sets;

[0075] A5: Confirm the writing of the object storage block to the client and complete the entire data writing process.

[0076] The local API processing layer and protocol layer call the heterogeneous distributed storage core layer to provide block storage, object storage and file storage services. The most important thing is the data reading and writing process. The reading and writing processes are roughly the same. Based on the above data writing process, the following describes the data reading process. If the library function reads a file through the client, it enters the file reading process. The file reading process includes:

[0077] B1: Get the index node from the file and calculate the primary storage data node (Primary OSD) where the index node is located based on the index node;

[0078] B2: Initiate a read request to the primary storage data node (Primary OSD). If the primary storage data node (Primary OSD) is damaged, send a read request to other storage data nodes (Secondary OSD and Tertiary OSD) redundant to the primary storage data node (OSD) according to priority until the object storage block (Object) is read;

[0079] B3: Return the read object storage block (Object) to the client (Client).

[0080] The virtualized storage method provided above has made some optimizations in the local API processing layer. When each storage data node (OSD) writes data into the memory buffer, it first sends a confirmation to the local API processing layer. When each storage data node (OSD) drops the data to the disk, it sends a final confirmation character (ACK) to the local API processing layer. Users can determine which confirmation to use to perform subsequent actions based on specific application scenarios. In short, without relying on other system modules, only the local API processing layer is needed to complete the addressing of the storage data node (OSD), so the local API processing layer can directly operate in parallel with the storage data node (OSD) and share the load as evenly as possible, thereby avoiding a single storage data node (OSD) from becoming a performance bottleneck.

[0081] It should be noted that any process or method description in the flowchart or otherwise described herein may be understood to represent a module, fragment or portion of a code comprising one or more executable instructions for implementing steps of a specific logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations, in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present application belong.

[0082] It should be understood that the various parts of the present application can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0083] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0084] In addition, each functional unit in each embodiment of the present application may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0085] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc.

[0086] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations on the present application. Ordinary technicians in this field can change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A virtualized storage pool based on heterogeneous instruction sets, characterized in that: The virtualized storage pool includes: The protocol layer is used to provide block storage, object storage and file storage protocols, and input data read and write commands through the block storage, object storage and file storage protocols; A local API processing layer, wherein the local API processing layer is used to provide library functions and execute data read and write commands by calling library functions; A heterogeneous distributed storage core layer, which provides storage data nodes with different instruction sets, and each storage data node establishes communication with the local API processing layer through the Ceph protocol, and is combined into a server by at least three storage data nodes, and each different server corresponds to storage read and write data; The heterogeneous distributed storage core layer includes: A client, wherein the client establishes a communication connection with a library function of a local API processing layer; An object storage device, wherein the object storage device comprises a plurality of storage data nodes, each of the storage data nodes is deployed on a server with a different instruction set, and a communication connection is established between each of the storage data nodes; A monitoring and maintenance device, wherein the monitoring and maintenance device establishes a communication connection with each of the storage data nodes and exchanges status information of each of the storage data nodes; The process of writing files in calling library functions to execute data read and write commands is as follows: A1: The client splits the file into several object storage blocks of the same size. A2: Map each object storage block to n mutually redundant storage data nodes, n ≥ 2; A3: Initiate a write request to the first-level storage data node among the n storage data nodes; A4: Synchronously replicate n-1 copies to n-1 storage data nodes with different instruction sets; A5: Confirm the writing of the object storage block to the client; The file reading process in calling the library function to execute the data read and write command is as follows: B1: Get the index node from the file and calculate the storage data node where the index node is located based on the index node; B2: initiates a read request to the storage data node; B3: Returns the read object storage block to the client.

2. The virtualized storage pool based on heterogeneous instruction sets according to claim 1, characterized in that: The block storage communication is connected with a distributed system interface; the object storage communication is connected with a bare metal server interface and / or a virtual machine interface; and the file storage protocol communication is connected with a POSIX file system interface.

3. The virtualized storage pool based on heterogeneous instruction sets according to claim 1, characterized in that: The servers with different instruction sets are: LoongArch server, alpha server, ARM server and x86 server.

4. A virtualized storage method based on heterogeneous instruction sets, characterized in that: The virtualized storage method is applied to the virtualized storage pool based on heterogeneous instruction sets as claimed in claim 1, and the virtualized storage method includes: S1: Call library functions to execute read and write commands through the user end, block storage, object storage, or file storage; S2: If the library function reads a file through the client, it enters the file reading process; if the library function writes a file through the client, it enters the file writing process; The write file includes: A1: The client splits the file into several object storage blocks of the same size. A2: Map each object storage block to n mutually redundant storage data nodes, n ≥ 2; A3: Initiate a write request to the first-level storage data node among the n storage data nodes; A4: Synchronously replicate n-1 copies to n-1 storage data nodes with different instruction sets; A5: Confirm the writing of the object storage block to the client; The reading file includes: B1: Get the index node from the file and calculate the storage data node where the index node is located based on the index node; B2: initiates a read request to the storage data node; B3: Returns the read object storage block to the client.

5. The virtualized storage method based on heterogeneous instruction sets according to claim 4, characterized in that: The A1 includes: A101: split the file into several object storage blocks; A102: Number several object storage blocks to generate heterogeneous OIDs.

6. The virtualized storage method based on heterogeneous instruction sets according to claim 4, characterized in that: The above A2 includes: A201: Map each object storage block to an intermediate layer SG, and organize the object storage blocks through the intermediate layer SG, so that a "one-to-many" mapping relationship is formed between the intermediate layer SG and the object storage blocks; A202: Mapping an intermediate layer SG to a plurality of storage data nodes by mapping the intermediate layer SG to the storage data nodes, so that the intermediate layer SG can be dynamically migrated to a combination of different storage data nodes.

7. The virtualized storage method based on heterogeneous instruction sets according to claim 6, characterized in that: In A202, when the intermediate layer SG is allocated to multiple storage data nodes, each storage data node is a server of a different instruction set, or a server of the same instruction set.

8. The virtualized storage method based on heterogeneous instruction sets according to claim 4, characterized in that: B2 is: initiating a read request to a storage data node. If the storage data node is damaged, sending a read request to other storage data nodes that are redundant to the storage data node according to priority until the object storage block is read.