Network io acceleration method, medium and product for shared storage database cluster

Abstract

The application provides a network IO acceleration method, medium and product of a shared storage database cluster. The network IO acceleration method of the shared storage database cluster comprises the following steps: a request node pre-allocates a local cache slot position identifier, and sends a lock application message carrying the local cache slot position identifier to a lock master node corresponding to a cache fusion distributed lock; the lock master node obtains a holding lock node and a holding lock end cache slot position identifier from the cache fusion distributed lock, and sends the request end cache slot position identifier and the holding lock end cache slot position identifier to the holding lock node; the holding lock node reads a target page in a buffer memory pool according to the holding lock end cache slot position identifier, and returns the page and the request end cache slot position identifier to the request node; and the request node stores the page in a corresponding position of the buffer memory pool according to the received request end cache slot position identifier. According to the scheme of the application, the cache slot position identifier is transmitted by relying on the cache fusion distributed lock, the hash retrieval is omitted, and the cluster IO efficiency and the overall throughput are effectively improved.

Description

Technical Field

[0001] This invention relates to the field of database technology, and in particular to a network I / O acceleration method, medium, and product for a shared storage database cluster. Background Technology

[0002] In a shared storage cluster environment, the sources of data pages (buffers) in a write-read-write cluster can generally be divided into three categories: disk reads, local cache hits, and remote node buffer hits. Remote node buffer hits are a crucial component of the shared storage architecture. The process involves accessing a data page and finding that it is not present in the local node's cache but is present in the caches of other nodes. The node holding the page then sends the data page to the requesting node; this read process is called a remote buffer hit.

[0003] Based on Oracle's Cache Fusion technology, each data page in a shared storage cluster has a unique Cache Fusion Lock. This is a distributed lock (Distributed Lock Manager, DLM) specifically used in shared storage database clusters. Any node that needs to access any data page must first acquire the Cache Fusion Lock corresponding to that page. The lock is allocated through a consistent hashing algorithm. Each Cache Fusion Lock corresponds to a unique lock master management node in the cluster. This node records information such as the lock-holding nodes and waiting queues for the managed Cache Fusion Lock.

[0004] In Real Application Clusters (RAC), requesting a page requires first requesting the page's DLM (Distributed Cache-Fusion Lock). The request first reaches the master node of the distributed cache lock, which forwards the request to the lock-holding node. The lock-holding node releases the lock and sends the data page to the requesting node. This is a complete network I / O operation in RAC.

[0005] Typically, processes that perform network I / O are hot processes, with multiple processes sending requests to them. The response speed of this hot process directly determines the overall throughput of the cluster. In the network I / O process of RAC, the process that performs network I / O is often the Global Cache Service Process (LMS). The LMS is responsible for many tasks such as sending and receiving messages, granting cache-integrated distributed locks, and sending data pages. Reducing the time spent in any of these steps can improve the overall throughput of the LMS, which is reflected in the increased throughput of the entire cluster.

[0006] For the data page transmission stage, LMS typically only needs to do two things: first, find the target data page, and second, send the data page to the node that requested it. Currently, the most common solution for finding the target data page is hash search, which ideally completes the search in O(1) time. However, hash search may have hash collisions, and it also often involves concurrency control, so the overall search overhead is longer than the optimal time. Summary of the Invention

[0007] One object of the present invention is to provide a network I / O acceleration method for shared storage database clusters that can solve any of the above problems.

[0008] A further objective of this invention is to improve the network I / O efficiency of remote page reading in a shared storage database cluster, thereby increasing the overall throughput of the cluster.

[0009] Another further objective of this invention is to address the problem that the lock-holding node incurs additional computational overhead by relying on the hash retrieval buffer memory pool, thus slowing down the cluster's I / O speed.

[0010] Specifically, this invention provides a network I / O acceleration method for a shared storage database cluster, wherein each node in the cluster has an array-based buffer memory pool, the array index of the buffer memory pool is a cache slot identifier, the cluster manages page caching through a cache-fused distributed lock, and the cache-fused distributed lock records a mapping between the lock-holding node and the cache slot identifier of the lock-holding end. The method includes: the requesting node pre-allocates a local cache slot identifier and sends a lock request message carrying the local cache slot identifier to the lock master node of the corresponding cache-fused distributed lock; the lock master node obtains the lock-holding node and the cache slot identifier of the lock-holding end from the cache-fused distributed lock, and sends the cache slot identifier of the request end and the cache slot identifier of the lock-holding end to the lock-holding node; the lock-holding node reads the target page in the buffer memory pool according to the cache slot identifier of the lock-holding end, and sends the page and the cache slot identifier of the request end back to the requesting node; and the requesting node stores the page in the corresponding position of the buffer memory pool according to the received cache slot identifier of the request end.

[0011] Optionally, before the step of pre-allocating local cache slot identifiers by the requesting node, the following steps are also included: the backend request process of the requesting node pre-allocates cache slot identifiers; and when sending a lock request message, the lock request message is forwarded to the lock master node through the local lock management service process; the lock management service processes of all nodes in the cluster uniformly undertake the tasks of forwarding lock messages and sending and receiving data pages, and encapsulate the cache slot identifiers and data pages into a unified message for transmission.

[0012] Optionally, while the lock-holding node reads the target page from the buffer memory pool based on the lock-holding end cache slot identifier, it also includes: the lock-holding node releasing the cache-fused distributed lock corresponding to the target page.

[0013] Optionally, after the step of storing the page in the corresponding location of the buffer memory pool, the process further includes: the requesting process sending an acknowledgment message carrying the cache slot identifier of the requesting end to the lock master node.

[0014] Optionally, after sending the confirmation message carrying the cache slot identifier of the requesting end, the method further includes: after receiving the confirmation message, the lock master node records the requesting node identifier, the cache slot identifier of the requesting end, and the current lock occupancy mode in the corresponding cache fusion distributed lock.

[0015] Optionally, the buffer memory pool is a contiguous array in the node's memory, where each cache slot identifier uniquely corresponds to an independent memory slot used for caching data pages.

[0016] Optionally, the cache integrates the distributed lock to store a one-to-one mapping relationship between the lock-holding node, the waiting queue node, and the cache slot identifier of each node.

[0017] Optionally, the buffer memory pool can be directly addressed by using the cache slot identifier of the request end and the cache slot identifier of the lock master node, eliminating the hash search process of the lock-holding node to retrieve the buffer memory pool and reducing the network I / O time of remote page reading in the cluster.

[0018] According to another aspect of the present invention, a computer-readable storage medium is also provided, on which a computer program is stored, wherein the computer program, when executed by a processor, implements the steps of the network I / O acceleration method for any of the above-described shared storage database clusters.

[0019] According to another aspect of the present invention, a computer program product is also provided, comprising a computer program, characterized in that, when executed by a processor, the computer program implements the steps of the network I / O acceleration method for any of the above-described shared storage database clusters.

[0020] The network I / O acceleration method for shared storage database clusters of the present invention utilizes a cache-fused distributed lock to record the mapping relationship of cache slot identifiers of each node. The lock master node relays the bidirectional cache slot identifiers to the lock-holding node, eliminating the hash search step of the lock-holding node in retrieving the buffer memory pool, significantly reducing the computation time of the cluster communication process, and shortening the network I / O processing time of a single remote cache page read.

[0021] Furthermore, the network I / O acceleration method for shared storage database clusters of the present invention can reduce the performance loss caused by hash collisions and concurrent retrieval, reduce the load of hot processes in the cluster, effectively improve the processing efficiency of remote data access in shared storage database clusters, and optimize the overall cluster read and write throughput.

[0022] The above and other objects, advantages and features of the present invention will become more apparent to those skilled in the art from the following detailed description of specific embodiments of the invention in conjunction with the accompanying drawings. Attached Figure Description

[0023] The following sections will describe some specific embodiments of the invention in detail by way of example and not limitation, with reference to the accompanying drawings. The same reference numerals in the drawings denote the same or similar parts or portions. Those skilled in the art should understand that these drawings are not necessarily drawn to scale. In the drawings: Figure 1 This is a schematic flowchart of a network I / O acceleration method for a shared storage database cluster according to an embodiment of the present invention; Figure 2 This is a schematic diagram of a computer program product according to an embodiment of the present invention; Figure 3 This is a schematic diagram of a computer-readable storage medium according to an embodiment of the present invention; and Figure 4 This is a schematic diagram of a computer device according to an embodiment of the present invention. Detailed Implementation

[0024] Those skilled in the art should understand that the embodiments described below are merely a part of the embodiments of the present invention, and not all of the embodiments of the present invention. These partial embodiments are intended to explain the technical principles of the present invention and are not intended to limit the scope of protection of the present invention. Based on the embodiments provided by the present invention, all other embodiments obtained by those skilled in the art without creative effort should still fall within the scope of protection of the present invention.

[0025] It should be noted that the logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus or device (such as a computer-based system, a processor-included system or other system that can fetch and execute instructions from, an instruction execution system, apparatus or device).

[0026] This embodiment provides a method for accelerating network I / O in a shared storage database cluster. It can utilize a cache-fused distributed lock to transmit cache slot identifiers, eliminating hash retrieval operations and significantly shortening the network I / O time for remote page reading. It also reduces the computational load on cluster nodes and effectively improves the data access efficiency and overall throughput of the shared storage database cluster. Figure 1 This is a schematic flowchart of a network I / O acceleration method for a shared storage database cluster according to an embodiment of the present invention, as follows: Figure 1 As shown, network I / O acceleration methods for shared storage database clusters generally include the following steps: Step S102: The requesting node pre-allocates a local cache slot identifier and sends a lock request message carrying the local cache slot identifier to the lock master node of the corresponding cache fusion distributed lock.

[0027] In step S104, the lock master node obtains the lock-holding node and the cache slot identifier of the lock-holding end from the cache fusion distributed lock, and sends the cache slot identifier of the requesting end and the cache slot identifier of the lock-holding end to the lock-holding node.

[0028] In step S106, the lock-holding node reads the target page from the buffer memory pool based on the lock-holding end cache slot identifier, and sends the page and the request end cache slot identifier back to the request node.

[0029] In step S108, the requesting node stores the page into the corresponding location in the buffer memory pool according to the cache slot identifier received from the requesting end.

[0030] Specifically, before step S102, which requests the node to pre-allocate a local cache slot identifier, the process may further include: pre-allocating a cache slot identifier by the backend request process of the request node.

[0031] Furthermore, when sending a lock request message, the lock request message is forwarded to the lock master node via the local lock management service process; the lock management service processes of all nodes in the cluster uniformly undertake the tasks of forwarding lock messages and sending and receiving data pages, and encapsulate the cache slot identifier and data page into a unified message for transmission.

[0032] Specifically, in step S106, while the lock-holding node reads the target page in the buffer memory pool according to the lock-holding end cache slot identifier, it may also include: the lock-holding node releasing the cache fusion distributed lock corresponding to the target page.

[0033] Specifically, after step S108, which stores the page in the corresponding location of the buffer memory pool, the process may further include: the requesting process sending an acknowledgment message carrying the cache slot identifier of the requesting end to the lock master node.

[0034] Specifically, after sending the confirmation message carrying the cache slot identifier of the requesting end, the process may further include: after receiving the confirmation message, the lock master node records the requesting node identifier, the cache slot identifier of the requesting end, and the current lock occupancy mode in the corresponding cache fusion distributed lock.

[0035] Specifically, the buffer memory pool is a contiguous array in the node's memory, with each cache slot uniquely corresponding to an independent memory slot used for caching data pages.

[0036] Specifically, the cache integrates the distributed lock to store the one-to-one mapping relationship between the lock-holding node, the waiting queue node, and the cache slot identifier of each node.

[0037] Specifically, the buffer memory pool is directly addressed by using the cache slot identifier of the request end and the cache slot identifier of the lock master node, which eliminates the hash search process of the lock-holding node to retrieve the buffer memory pool and reduces the network I / O time of remote page reading in the cluster.

[0038] It should be noted that the network I / O acceleration method for the shared storage database cluster in this embodiment utilizes a cache-fused distributed lock, allowing the master node of the cache-fused distributed lock to know the location of the data page corresponding to the cache-fused distributed lock managed by this node on all nodes in the cluster. During the network I / O process, the master node will directly send the location of the data page on the lock-holding node, thereby avoiding the lock-holding node from performing a hash search. This saves the overhead of a hash search step in the communication process LMS.

[0039] Specifically, the network I / O acceleration method for the shared storage database cluster in this embodiment can include the following as a whole: Data pages are arranged in memory as an array called the BufferPool. Each data page has a buffer slot identifier (bufferid) in memory, which is the index of the BufferPool array.

[0040] Before a requesting node requests a data page from the lock master node, it will first allocate a cache slot identifier. After the requesting node obtains the data page, the data page will be placed in the buffer memory pool at the [cache slot identifier] location.

[0041] The requesting node sends a request to the lock master node, and the message carries the cache slot identifier of the requesting node to the lock master node.

[0042] The lock master node parses the request, obtains the current lock-holding node of the cache-fused distributed lock corresponding to the data page requested by the requesting node, and finds the cache slot identifier of the lock-holding node stored in the cache-fused distributed lock.

[0043] The master node will send the cache slot identifier of the requesting node and the cache slot identifier of the lock-holding node together to the lock-holding node.

[0044] The lock-holding node receives the message sent by the lock master node, parses the lock-holding cache slot identifier carried in it, releases the cache fusion distributed lock held by this node, and reads the data page based on the lock-holding node's cache slot identifier. It then packages the data page and the requesting node's cache slot identifier together and sends them to the requesting node.

[0045] The requesting node will receive the cache slot identifier of the requesting node and the requested data page from the locking node. After that, the process that receives the data page on the requesting node only needs to place the received data page into the position of the received cache slot identifier. After the data page is received, the process that requested the data page will be woken up.

[0046] After obtaining the data page, the process requesting the data page needs to send an acknowledgment message to the master node of the cache-integrated distributed lock. This message will carry the cache slot identifier of the requesting node. After that, the requesting process will no longer sleep and will continue its subsequent work normally.

[0047] After receiving the acknowledgment message from the requesting node, the lock master node parses the message and extracts the cache slot identifier of the requesting node. At the same time, it leaves a record in the cache-fused distributed lock, including the lock mode requested by the requesting node and the cache slot identifier of the requesting node.

[0048] It should be emphasized that the network I / O acceleration method for the shared storage database cluster in this embodiment has the following key technical points: The cache-integrated distributed lock contains a cache slot identifier structure to identify the data page location of each node.

[0049] Before initiating a cache fusion distributed lock request, the process requesting the cache fusion distributed lock needs to pre-allocate a space for the data page.

[0050] The requesting node needs to send the requested data page location along with the cached distributed lock request to the lock master node.

[0051] The master node reads the lock-holding node and the corresponding cache slot identifier recorded in the cache fusion distributed lock, and sends the cache slot identifiers of both parties to the lock-holding node.

[0052] The lock-holding node receives the message sent by the lock master node, reads the corresponding data page according to the cache slot identifier in the message, and sends the data page together with the data page location of the requesting node.

[0053] The requesting node will receive the data page requested by the locking node and the pre-allocated data page position. The requesting node will then place the received data page into the pre-allocated slot.

[0054] After receiving the data page, the requesting node sends the pre-allocated cache slot information to the lock master node via an acknowledgment message. After this, the requesting node no longer waits.

[0055] After receiving the confirmation message, the lock-holding node will update the lock information of the requesting node, mainly updating the data page slot information of the requesting node.

[0056] In summary, the network I / O acceleration method for the shared storage database cluster in this embodiment utilizes a cache-fused distributed lock to record the mapping relationship of cache slot identifiers for each node. The lock master node relays the bidirectional cache slot identifiers to the lock-holding node, eliminating the hash search step of the lock-holding node in retrieving the buffer memory pool. This significantly reduces the computation time of the cluster communication process and shortens the network I / O processing time for a single remote cache page read.

[0057] Furthermore, the network I / O acceleration method for the shared storage database cluster in this embodiment can reduce the performance loss caused by hash collisions and concurrent retrieval, reduce the load on hot processes in the cluster, effectively improve the processing efficiency of remote data access in the shared storage database cluster, and optimize the overall cluster read and write throughput.

[0058] The flowcharts provided in the above embodiments are not intended to indicate that the operations of the method will be performed in any particular order, or that all operations of the method are included in all every case. Furthermore, the method may include additional operations. Within the scope of the technical concept provided by the methods in the above embodiments, additional variations can be made to the above methods.

[0059] This embodiment also provides a computer program product, a computer-readable storage medium, and a computer device. Figure 2 This is a schematic diagram of a computer program product 500 according to an embodiment of the present invention. Figure 3 This is a schematic diagram of a computer-readable storage medium 300 according to an embodiment of the present invention. Figure 4 This is a schematic diagram of a computer device 400 according to an embodiment of the present invention.

[0060] Computer program product 500 includes computer program 310, which, when executed by processor 410, implements the steps of the network I / O acceleration method for the shared storage database cluster described above. Computer-readable storage medium 300 stores the aforementioned computer program 310, which, when executed by processor 410, implements the steps of the network I / O acceleration method for the shared storage database cluster described above. Computer device 400 may include memory 420, processor 410, and computer program 310 stored in memory 420 and running on processor 410.

[0061] The computer program 310 used to perform the operations of the present invention may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-related instructions, microcode, firmware instructions, status setting data, integrated circuit configuration data, or source code or object code written in any combination of one or more programming languages ​​and procedural programming languages.

[0062] Computer program 310 may execute entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the latter case, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0063] In some embodiments, in order to perform aspects of the present invention, electronic circuits including, for example, programmable logic circuits, field-programmable gate arrays (FPGAs) or programmable logic arrays (PLAs) can execute computer-readable program instructions to personalize the electronic circuits by utilizing state information of computer-readable program instructions.

[0064] For the purposes of this embodiment, computer program product 500 is a related product containing computer program 310. For the purposes of this embodiment, computer-readable storage medium 300 is a tangible device capable of holding and storing computer program 310, and can be any device capable of containing, storing, communicating, propagating or transmitting computer program 310 for use by or in conjunction with an instruction execution system, apparatus or device.

[0065] More specific examples (a non-exhaustive list) of computer-readable storage media 300 include the following: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), portable optical disc read-only memory (CD-ROM), digital multifunction disc (DVD), memory sticks, floppy disks, mechanical encoding devices, and any suitable combination of the foregoing.

[0066] Computer device 400 may include memory 420, processor 410, and computer program 310 stored on memory 420 and running on processor 410, wherein processor 410 executes computer program 310 to implement the steps of the network I / O acceleration method for shared storage database cluster of any of the above embodiments.

[0067] It should be noted that the logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be specifically implemented in any machine-readable storage medium for use by, or in conjunction with, instruction execution systems, apparatuses or devices (such as computer-based systems, processor-based systems or other systems that can fetch and execute instructions from, or instruction execution systems, apparatuses or devices).

[0068] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system.

[0069] Computer device 400 can be, for example, a server, desktop computer, laptop computer, tablet computer, or smartphone. In some examples, computer device 400 can be a cloud computing node. Computer device 400 can be described in the general context of computer system executable instructions (such as program modules) executed by a computer system. Typically, program modules can include routines, programs, object programs, components, logic, data structures, etc., that perform specific tasks or implement specific abstract data types. Computer device 400 can be implemented in a distributed cloud computing environment where tasks are performed by remote processing devices linked through a communication network. In a distributed cloud computing environment, program modules can reside on local or remote computing system storage media, including storage devices.

[0070] Computer device 400 may include a processor 410 adapted to execute stored instructions and a memory 420 that provides temporary storage space for the operation of instructions during operation. The processor 410 may be a single-core processor, a multi-core processor, a computing cluster, or any other configuration. The memory 420 may include random access memory (RAM), read-only memory, flash memory, or any other suitable storage system.

[0071] The processor 410 can be connected via a system interconnect (e.g., PCI, PCI-Express, etc.) to an I / O interface (input / output interface) suitable for connecting the computer device 400 to one or more I / O devices (input / output devices). I / O devices may include, for example, a keyboard and indicating devices, where indicating devices may include a touchpad or touchscreen, etc. I / O devices may be built into the computer device 400 or may be external devices connected to the computing device.

[0072] The processor 410 may also be linked via a system interconnect to a display interface suitable for connecting the computer device 400 to a display device. The display device may include a display screen as a built-in component of the computer device 400. The display device may also include an external computer monitor, television, or projector connected to the computer device 400. Furthermore, a network interface controller (NIC) may be adapted to connect the computer device 400 to a network via a system interconnect. In some embodiments, the NIC may use any suitable interface or protocol (such as an Internet Minicomputer System Interface) to transmit data. The network may be a cellular network, a radio network, a wide area network (WAN), a local area network (LAN), or the Internet, etc. Remote devices may connect to the computing device via the network.

[0073] Therefore, those skilled in the art should recognize that although numerous exemplary embodiments of the present invention have been shown and described in detail herein, many other variations or modifications conforming to the principles of the present invention can be directly determined or derived from the disclosure of the present invention without departing from the spirit and scope of the invention. Thus, the scope of the present invention should be understood and construed as covering all such other variations or modifications.

Claims

1. A method for accelerating network I / O in a shared storage database cluster, wherein each node of the cluster is provided with an array-based buffer memory pool, the array index of the buffer memory pool is a cache slot identifier, the cluster manages page caching through a cache-fused distributed lock, and the cache-fused distributed lock records a mapping between the lock-holding node and the cache slot identifier of the lock-holding end, the method comprising: The requesting node pre-allocates a local cache slot identifier and sends a lock request message carrying the local cache slot identifier to the lock master node of the corresponding cache fusion distributed lock; The lock master node obtains the lock-holding node and the lock-holding end cache slot identifier from the cache fusion distributed lock, and sends the request end cache slot identifier and the lock-holding end cache slot identifier to the lock-holding node; The lock-holding node reads the target page from the buffer memory pool based on the lock-holding end cache slot identifier, and sends the page and the request end cache slot identifier back to the request node; as well as The requesting node stores the page into the corresponding location in the buffer memory pool based on the received cache slot identifier from the requesting end.

2. The method of claim 1, further comprising, prior to the step of pre-allocating local cache slot identifiers on the request node: The cache slot identifier is pre-allocated by the backend request process of the request node; and When the lock request message is sent, the lock request message is forwarded to the lock master node through the local lock management service process; the lock management service processes of all nodes in the cluster uniformly undertake the tasks of lock message forwarding and data page sending and receiving, and encapsulate the cache slot identifier and the data page into a unified message for transmission.

3. The method according to claim 1, wherein while the lock-holding node reads the target page from the buffer memory pool based on the lock-holding end cache slot identifier, it further includes: The lock-holding node releases the cache-fused distributed lock corresponding to the target page.

4. The method according to claim 1, further comprising, after the step of storing the page in the corresponding location of the buffer memory pool: The requesting process sends an acknowledgment message carrying the cache slot identifier of the requesting end to the lock master node.

5. The method of claim 4, further comprising, after the step of sending an acknowledgment message carrying the identifier of the requesting end cache slot: After receiving the confirmation message, the lock master node records the requesting node identifier, the requesting cache slot identifier, and the current lock occupancy mode in the corresponding cache fusion distributed lock.

6. The method according to claim 1, wherein, The buffer memory pool is an array arranged continuously in the node memory, and each cache slot identifier uniquely corresponds to an independent memory slot used for caching data pages.

7. The method according to claim 1, wherein, The cache-fused distributed lock stores a one-to-one mapping relationship between the lock-holding node, the waiting queue node, and the cache slot identifier of each node.

8. The method according to claim 1, wherein, By relaying the cache slot identifiers of the requesting end and the cache slot identifiers of the lock-holding end through the lock master node, the buffer memory pool can be directly addressed, eliminating the hash search process for the lock-holding node to retrieve the buffer memory pool and reducing the network I / O time consumption for remote page reading of the cluster.

9. A computer-readable storage medium having a computer program stored thereon, the computer program being executed by a processor to implement the steps of the network I / O acceleration method for a shared storage database cluster according to any one of claims 1 to 8.

10. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the steps of the network I / O acceleration method for a shared storage database cluster as described in any one of claims 1 to 8.

Images