WO2021164163A1

WO2021164163A1 - Request processing method and apparatus, device and storage medium

Info

Publication number: WO2021164163A1
Application number: PCT/CN2020/097969
Authority: WO
Inventors: 李宏伟; 张东
Original assignee: 苏州浪潮智能科技有限公司
Priority date: 2020-02-21
Filing date: 2020-06-24
Publication date: 2021-08-26
Also published as: CN111371848A

Abstract

An IO scheduling layer based request processing method and apparatus, a device and a storage medium. In the solution, for SCSI requests sent by an application layer, if the size of data to be processed corresponding to the SCSI requests does not exceed a preset threshold, the SCSI requests are not sent firstly to a software stack but the IO scheduling layer; and the received requests are merged by means of the IO scheduling layer and then sent to the software stack, so that the requests are sent to corresponding server nodes by means of a message routing system. The mode, by which the SCSI requests are processed by means of the IO scheduling layer and then sent to the software stack, can reduce, by merging the requests, the number of times of interactions passing through the software stack, thus improving throughput capacity of a storage system.

Description

Request processing method, device, equipment and storage medium

This application claims the priority of a Chinese patent application filed with the Chinese Patent Office on February 21, 2020, the application number is 202010108904.1, and the invention title is "a request processing method, device, equipment and storage medium", the entire content of which is incorporated by reference Incorporated in this application.

Technical field

The present invention relates to the field of distributed storage, and more specifically, to a request processing method, device, equipment and computer-readable storage medium based on a distributed storage system.

Background technique

A distributed storage system is a cluster composed of a large number of standard servers, which provides storage functions as a whole to the outside world, and the data is actually distributed on each server, which is essentially a software-defined storage. In terms of the realization of the iscsi block storage function provided by the specific distributed storage system, the SCSI command received by the iscsi target software stack is generally unpacked directly, and then passed to the server node where the data is located through the message routing system for processing, and then The data is returned to the upper application via the routing system. It can be seen that for a distributed system, it consists of multiple layers of software stacks, and each SCSI request needs to span multiple software stacks or even across hosts, so the processing of a single SCSI request is slower than traditional centralized storage.

Summary of the invention

The purpose of the present invention is to provide a request processing method, device, equipment and computer-readable storage medium based on a distributed storage system to improve the processing speed of SCSI requests.

To achieve the above objective, the present invention provides a request processing method based on a distributed storage system. Based on the IO scheduling layer, the request processing method includes:

Receive the SCSI request sent by the application layer;

If the size of the to-be-processed data corresponding to the SCSI request does not exceed the predetermined threshold, use the scheduling strategy to analyze the received SCSI request, perform request merging processing, and encapsulate the processed requests separately;

The encapsulated request is sent to the software stack, so that the software stack sends the encapsulated request to the corresponding server node through the message routing system.

Wherein, the receiving the SCSI request sent by the application layer includes:

The SCSI request sent by the application layer is received through the scheduling queue of the IO scheduling layer.

Wherein, the IO scheduling layer includes an IO scheduling interface, and the IO scheduling interface accesses different scheduling strategies in the form of a plug-in.

Wherein, the use of the scheduling strategy to analyze the received SCSI request includes:

The scheduling strategy is used to analyze the received SCSI requests in the current time period.

Wherein, if the size of the data to be processed corresponding to the SCSI request exceeds a predetermined threshold, the request processing method further includes:

Judging whether the to-be-processed data corresponding to the SCSI request is multi-copy data;

If so, the SCSI request is split into multiple SCSI sub-requests according to the number of copies of the data to be processed, and each SCSI sub-request is encapsulated and sent to the software stack, so that the software stack passes the encapsulated request through the message The routing system sends to the server node corresponding to each SCSI sub-request.

Wherein, the use of the scheduling strategy to analyze the received SCSI request and perform request merging processing includes:

If there are at least two SCSI requests that are connected in the first digit of the address, merge the SCSI requests connected in the first digit of the address into one request;

If the storage objects corresponding to at least two SCSI requests are the same storage object, the SCSI requests of the same storage object are combined into one request;

If the server nodes corresponding to at least two SCSI requests are the same server node, the SCSI requests of the same server node are combined into one request.

To achieve the above objective, the present invention further provides a request processing device based on a distributed storage system. Based on the IO scheduling layer, the request processing device includes:

The receiving module is used to receive the SCSI request sent by the application layer;

The parsing module is used for parsing the received SCSI request using a scheduling strategy when the size of the data to be processed corresponding to the SCSI request does not exceed a predetermined threshold, and performing request merging processing;

Encapsulation module, used to encapsulate the processed request separately;

The sending module is used to send the encapsulated request to the software stack, so that the software stack sends the encapsulated request to the corresponding server node through the message routing system.

Among them, this program also includes:

A judging module, configured to judge whether the to-be-processed data corresponding to the SCSI request is multi-copy data when the size of the to-be-processed data corresponding to the SCSI request exceeds a predetermined threshold;

A splitting module, configured to split the SCSI request into multiple SCSI sub-requests according to the number of copies of the to-be-processed data when the to-be-processed data corresponding to the SCSI request is multiple copies of data;

The encapsulation module is used to encapsulate each SCSI sub-request;

The sending module is used to send each encapsulated SCSI sub-request to the software stack, so that the software stack sends the encapsulated request to the server node corresponding to each SCSI sub-request through the message routing system. To achieve the above objective, the present invention further provides a request processing device based on a distributed storage system, including:

Memory, used to store computer programs;

The processor is used to implement the steps of the above request processing method when the computer program is executed.

In order to achieve the foregoing objective, the present invention further provides a computer-readable storage medium having a computer program stored on the computer-readable storage medium, and when the computer program is executed by a processor, the steps of the foregoing request processing method are implemented.

It can be seen from the above solution that a request processing method based on a distributed storage system provided by an embodiment of the present invention is based on the IO scheduling layer. The request processing method includes: receiving a SCSI request sent by the application layer; determining the pending request corresponding to the SCSI request. Whether the processing data size exceeds a predetermined threshold; if not, use the scheduling strategy to analyze the received SCSI request, perform request merging processing, and encapsulate the processed requests separately; send the encapsulated request to the software stack to make The software stack sends the encapsulated request to the corresponding server node through the message routing system.

It can be seen that in this application, the SCSI request sent by the application layer is not sent to the software stack, but first sent to the IO scheduling layer, and the received requests are merged through the IO scheduling layer, and then sent to the software stack. It can be sent to the corresponding server node through the message routing system; this method of processing the SCSI request through the IO scheduling layer and then sending it to the software stack can reduce the number of interactions through the software stack by combining requests and improve the storage system Throughput.

The present invention also discloses a request processing device, equipment and computer-readable storage medium based on the distributed storage system, which can also achieve the above technical effects.

Description of the drawings

In order to explain the embodiments of the present invention or the technical solutions in the prior art more clearly, the following will briefly introduce the drawings that need to be used in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only These are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative work.

FIG. 1 is a schematic flowchart of a request processing method based on a distributed storage system disclosed in an embodiment of the present invention;

2 is a schematic structural diagram of a request processing apparatus based on a distributed storage system disclosed in an embodiment of the present invention;

Fig. 3 is a schematic structural diagram of a request processing device based on a distributed storage system disclosed in an embodiment of the present invention.

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative work shall fall within the protection scope of the present invention.

It is understandable that a distributed storage system is a cluster composed of a large number of standard servers. In terms of implementation, the distributed storage system can be subdivided into distributed file systems, distributed object storage systems, and distributed block systems, including Distributed database, etc. Compared with traditional centralized storage, distributed storage has the advantages of low cost, strong scalability, easy use and management, and no single point of failure. Distributed storage system architecture is the core of HCI (Hyper Converged Infrastructure). A commonly used solution is to use distributed storage system as a software service, and provide block devices for upper hypervisors through RBD or iSCSI protocols. Store virtual machine data and other massive data. In terms of the realization of the iscsi block storage function provided by the specific distributed storage system, the SCSI command received by the iscsi target software stack is generally unpacked directly, and then passed to the server node where the data is located through the message routing system for processing, and then The data is returned to the upper application via the routing system.

The traditional centralized storage for processing requests for SCSI commands is processed by an embedded processor set integrated with the hard disk. It has the characteristics of simple logic but block speed, so it is suitable for direct processing of a single SCSI command. For a distributed system, it is composed of multiple layers of software stacks, and each SCSI request needs to span multiple software stacks or even across the host, so the processing of a single SCSI request is slower than traditional centralized storage. Specifically, based on the principle of "high cohesion and low coupling" in the basic software rules, software stacks are generally almost independent, have their own independent workflows, and are usually asynchronous. Therefore, a request is transmitted between different software stacks. When processing: 1. It needs to be packaged/parsed according to the agreed format; 2. The message needs to be transmitted through socket or rpc; 3. After reaching the opposite end, it is usually placed in the processing queue and needs to be queued for processing. These factors will increase processing overhead and latency.

However, the software-defined nature of the distributed storage system brings the advantage of flexibility. In response to the above problems, this solution discloses a request processing method, device, equipment and computer-readable storage medium based on the distributed storage system. The IO scheduling layer schedules SCSI requests from upper-layer applications. For example, by merging message processing, the number of interactions between different software stacks can be reduced, the processing speed of SCSI requests can be improved, and the delay caused by request calls across the software stack can be reduced to the maximum Optimize the throughput of the storage system.

Referring to FIG. 1, an embodiment of the present invention provides a request processing method based on a distributed storage system, and the request processing method includes:

S101: Receive a SCSI request sent by the application layer;

Wherein, the receiving the SCSI request sent by the application layer includes: receiving the SCSI request sent by the application layer through the scheduling queue of the IO scheduling layer. The IO scheduling layer includes an IO scheduling interface, and the IO scheduling interface accesses different scheduling strategies in the form of a plug-in.

It should be noted that in this application, an IO scheduling layer is added to the iscsi target software stack of the distributed storage system. In the existing solution, SCSI requests from the application layer are placed in the waiting queue, and then the IO processing thread As a consumer, take out the original SCSI message from the waiting queue and parse it. In order to improve the request processing speed, the present invention adds an IO scheduling layer. Specifically, the IO scheduling layer first needs to receive the SCSI request sent by the application layer, and first put the received SCSI request into the scheduling queue, and use the SCSI request as input. , In order to combine the original SCSI request scheduling and encapsulation according to the preset scheduling strategy, and then put it into the waiting queue of the software stack, hand it over to the IO processing thread, and then pass it to the storage node for processing via the message routing system.

S102. If the size of the to-be-processed data corresponding to the SCSI request does not exceed a predetermined threshold, use a scheduling strategy to parse the received SCSI request, perform request merging processing, and encapsulate the processed requests separately;

Wherein, using the scheduling strategy to analyze the received SCSI request includes: using the scheduling strategy to analyze the received SCSI request in the current time period.

In this application, corresponding auxiliary functions can be added to the IO scheduling layer, such as timers, monitoring interfaces, etc., specifically: the timer is used to control the merge window, which can be understood as the merge cycle, that is: This application records the merging cycle through a timer, and the current time period is the current merging cycle. This application will accumulate SCSI requests in the current time period and analyze the SCSI requests received in the current time period. For example: the merge period is 1 hour, 1:00～2:00 is a merge period, 2:00～3:00 is another merge period, if the current time is 1:30, the current time period is 1:00 This is a consolidation period of ~2:00, so when analyzing and consolidating SCSI requests, it is necessary to analyze the SCSI requests received within the current time period from 1:00 to 2:00. The time period can be customized and modified according to requirements, so that when the IO delay is met, more requests can be merged by increasing the time period. If the timer expires, the current merged request will be forced down. The monitoring interface can be used as an independent thread, which monitors the IO pressure of each node and runs at all times. When disassembling multi-copy requests, the disassembling request can be sent to the server node with lower IO pressure first, so as to realize the load balance of each server node. It can be seen that this application can avoid starvation by adding corresponding auxiliary functions in the IO scheduling layer, and ensure that all SCSI requests can return processing results within the user's specified time.

It should be noted that, in order to improve the flexibility of the IO scheduling layer and enable the distributed storage system to adopt appropriate scheduling strategies for different application scenarios, this solution adds a new feature to the IO scheduling framework based on the principle of separation of mechanisms and strategies. Group abstract interfaces, which are implemented in the form of plug-ins to access different scheduling strategies for targeted tuning. Specifically, the abstract interface design of the IO scheduling strategy in this solution mainly includes the following steps:

1. The io_sched_type structure is designed, which mainly contains interfaces related to specific scheduling strategies;

2. Realize the specific scheduling strategy algorithm, that is, realize the specific scheduling function according to the above interface, such as the IO request mode or disk type for different applications.

Specifically, if the disk type is a solid state drive SSD, because its processing speed is fast, the small block request does not need to perform the merge operation at this time. If the disk type is a traditional disk, the processing speed is slow, then this solution can be used to request Consolidation; further, if the IO request mode requires speed, the time period can be reduced, and if the IO request mode requires throughput, the time period can be increased; specifically, the IO mode mainly includes: large block request/small block request, direct /non-direct IO request, buffer io mode. For the small block request mode, the IO scheduling layer will merge as much as possible while meeting the delay, which can increase the storage throughput. For the bulk request mode, if the requested data is multiple copies of data, it can be split to different copy nodes to respond at the same time to improve read performance. For direct io mode, the request will be responded as quickly as possible, while for non-direct io mode, it will be processed normally; for buffer io mode, it will be combined as much as possible to improve throughput when the delay is met. Therefore, when analyzing the received SCSI request by using the scheduling strategy in this application, it can also be configured according to the different IO modes.

S103. Send the encapsulated request to the software stack, so that the software stack sends the encapsulated request to the corresponding server node through the message routing system.

It should be noted that the main purpose of the IO scheduling layer in this application is to merge messages and optimize the message distribution process. Under the condition that the delay is satisfied, the message processing is combined as much as possible to reduce the number of interactions between different software stacks. For example: For a request A, the time it takes to go through all the software stacks is t1, and the message processing time is t2; if it is not merged, the processing time for N requests is N*(t1+t2). If these messages can be merged, Then the processing time is approximately t1+N*T2. In distributed storage, for the processing of small blocks of IO, the overhead t1 introduced by software stack interaction is much greater than the time t2 for specific IO processing. In this case, if small blocks of IO can be combined, optimization is of great significance.

It can be seen that this application designs and implements the IO scheduling layer, and uses the IO scheduling layer as a consumer to take the SCSI requests of the application layer into the scheduling queue, and then analyze the SCSI requests in the scheduling queue according to the specified scheduling strategy. According to the node distribution of distributed storage and the characteristics of object storage, the merging process is performed, and then the encapsulated message is put into the waiting queue, handed over to the IO processing thread, and then sent to the corresponding node for processing through message routing. In this way, the performance and throughput of the distributed storage system can be greatly improved, and the IO mode of the distributed storage system can be adjusted for different application scenarios to improve product performance and competitiveness.

Based on the foregoing embodiment, in this embodiment, if the size of the data to be processed corresponding to the SCSI request exceeds a predetermined threshold, the request processing method further includes:

It should be noted that since data in distributed storage always has multiple copies, the IO scheduling layer in this application can not only merge SCSI requests, but also perform disassembly processing, that is, when responding to IO requests, If the size of the to-be-processed data corresponding to the SCSI request exceeds the predetermined threshold, and there are multiple copies of the to-be-processed data in the system, then the SCSI request can be split according to the number of copies of the One request is responsible for a piece of data to be processed, and the split request is sent to different servers that store the copy data at the same time, so that multiple servers respond at the same time, thereby increasing the reading rate.

Based on the above embodiment, in this embodiment, the scheduling strategy is used to parse the received SCSI request and perform the request merging process. Specifically, the process may include: if there are at least two SCSI requests corresponding to the address first connected, then the address The first connected SCSI requests are combined into one request; if there are at least two SCSI requests corresponding to the same storage object, the SCSI requests of the same storage object are combined into one request; if there are at least two SCSI requests corresponding to the same storage object If the server nodes are the same server node, the SCSI requests of the same server node are combined into one request.

In this application, the merger processing process mainly includes the following situations:

1. If the two request addresses are connected in the first place, then the two requests can be combined into one continuous request;

2. Two requests correspond to the same storage object at the bottom, then the two requests can be combined into the same request, and two non-adjacent data vectors are filled at the same time from the bottom;

3. If two requests are sent to the same node, then the two requests can be packaged and sent to the target storage node at once.

It should be noted that this application uses the IO scheduling layer to perform merging and disassembling of SCSI requests, which can redesign the IO processing unit consumed by the IO thread and the communication protocol of the distributed node. For example, in the original solution, the IO thread processes the communication protocol. The IO unit is a direct encapsulation of the SCSI request, that is, the SCSI request is the unit; in this application, the IO processing unit is the combined/split SCSI request, for example: SCSI requests A and B are connected in the first place of the address , Then the two SCSI requests will be combined into one message and sent to the bottom layer for processing, reducing the number of interactions; and, in the original scheme, after the node receives the request, it needs to determine whether the requested data copy is on the node, if not This node needs to be forwarded to the corresponding node. In this application, the node where the data copy is located will be determined before the request is distributed, so as to split the request to send to different nodes, so the node does not need to judge after receiving the request, and directly responds to the request That is, it not only simplifies the processing logic, but also improves the efficiency of data processing. It can be seen that the new IO processing unit provided by this application has been optimized for the characteristics of distributed storage, which can simplify the underlying processing and interaction logic.

In summary, the present invention proposes a design and implementation of an IO scheduling framework for a distributed storage system in the field of distributed storage. The IO scheduling layer is added to the iscsi target software stack of the distributed storage system to request SCSI As input, the request after scheduling is encapsulated and passed to the storage node through the message routing system for processing; a set of IO scheduling interfaces is designed and abstracted for the IO scheduling layer, and different IO scheduling strategies are implemented in the form of plug-ins. For different storage application scenarios. This solution can greatly improve the performance and throughput of the distributed storage system and improve product competitiveness.

The request processing apparatus provided by the embodiment of the present invention will be introduced below. The request processing apparatus described below and the request processing method described above may refer to each other.

Referring to FIG. 2, an embodiment of the present invention provides a request processing device based on a distributed storage system, based on the IO scheduling layer, and the request processing device includes:

The receiving module 100 is used to receive the SCSI request sent by the application layer;

The parsing module 200 is configured to analyze the received SCSI request by using a scheduling strategy when the size of the to-be-processed data corresponding to the SCSI request does not exceed a predetermined threshold, and perform request merging processing;

The encapsulation module 300 is used to encapsulate the processed requests separately;

The sending module 400 is configured to send the encapsulated request to the software stack, so that the software stack sends the encapsulated request to the corresponding server node through the message routing system.

Wherein, the receiving module is specifically configured to receive the SCSI request sent by the application layer through the scheduling queue of the IO scheduling layer. Wherein, the IO scheduling layer includes an IO scheduling interface, and the IO scheduling interface accesses different scheduling strategies in the form of a plug-in.

Wherein, the analysis module is specifically configured to analyze the received SCSI requests in the current time period by using a scheduling strategy.

Among them, this program also includes:

The encapsulation module is used to encapsulate each SCSI sub-request;

The sending module is used to send each encapsulated SCSI sub-request to the software stack, so that the software stack sends the encapsulated request to the server node corresponding to each SCSI sub-request through the message routing system.

Wherein, the parsing module is specifically configured to: when there are at least two SCSI requests corresponding to the address first connected, combine the SCSI requests connected at the first address into one request; there are at least two SCSI requests corresponding to the same storage object as the storage object When the SCSI requests of the same storage object are combined into one request; when there are at least two SCSI requests corresponding to the same server node, the SCSI requests of the same server node are combined into one request.

Referring to Figure 3, the embodiment of the present invention also discloses a schematic structural diagram of a request processing device based on a distributed storage system; the device specifically includes:

The memory 11 is used to store computer programs;

The processor 12 is configured to implement the steps of the request processing method described in any of the foregoing method embodiments when executing the computer program.

In this embodiment, the device may be a PC (Personal Computer, personal computer), or a terminal device such as a smart phone, a tablet computer, a palmtop computer, and a portable computer.

The device may include a memory 11, a processor 12, and a bus 13.

The memory 11 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), magnetic memory, magnetic disk, optical disk, and the like. In some embodiments, the memory 11 may be an internal storage unit of the device, such as a hard disk of the device. In other embodiments, the memory 11 may also be an external storage device of the device, such as a plug-in hard disk equipped on the device, a Smart Media Card (SMC), a Secure Digital (SD) card, and a flash memory card. (Flash Card) and so on. Further, the memory 11 may also include both an internal storage unit of the device and an external storage device. The memory 11 can be used not only to store application software and various types of data installed in the device, such as codes for executing a request processing method, etc., but also to temporarily store data that has been output or will be output.

The processor 12 may be a central processing unit (CPU), controller, microcontroller, microprocessor or other data processing chip in some embodiments, and is used to run the program code or processing stored in the memory 11 Data, such as code that executes the request processing method, etc.

The bus 13 may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus. The bus can be divided into address bus, data bus, control bus and so on. For ease of presentation, only a thick line is used in FIG. 3 to represent it, but it does not mean that there is only one bus or one type of bus.

Further, the device may also include a network interface 14. The network interface 14 may optionally include a wired interface and/or a wireless interface (such as a WI-FI interface, a Bluetooth interface, etc.), which is usually used to communicate between the device and other electronic devices. Establish a communication connection.

Optionally, the device may also include a user interface. The user interface may include a display (Display) and an input unit such as a keyboard (Keyboard). The optional user interface may also include a standard wired interface and a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode, organic light-emitting diode) touch device, etc. Among them, the display can also be appropriately called a display screen or a display unit, which is used to display the information processed in the device and to display a visualized user interface.

Figure 3 only shows a device with components 11-14. Those skilled in the art can understand that the structure shown in Figure 3 does not constitute a limitation on the device, and may include fewer or more components than shown. Or some parts are combined, or different parts are arranged.

The embodiment of the present invention also discloses a computer-readable storage medium having a computer program stored on the computer-readable storage medium, and when the computer program is executed by a processor, the request processing method described in any of the foregoing method embodiments is implemented A step of.

Among them, the storage medium may include: U disk, mobile hard disk, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic disk or optical disk, etc., which can store program code medium.

The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments, and the same or similar parts between the various embodiments can be referred to each other.

The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown in this document, but should conform to the widest scope consistent with the principles and novel features disclosed in this document.

Claims

A request processing method based on a distributed storage system, characterized in that, based on an IO scheduling layer, the request processing method includes:

Receive the SCSI request sent by the application layer;

If the size of the to-be-processed data corresponding to the SCSI request does not exceed the predetermined threshold, use the scheduling strategy to analyze the received SCSI request, perform request merging processing, and encapsulate the processed requests separately;

The encapsulated request is sent to the software stack, so that the software stack sends the encapsulated request to the corresponding server node through the message routing system.
The request processing method according to claim 1, wherein the receiving the SCSI request sent by the application layer comprises:

The SCSI request sent by the application layer is received through the scheduling queue of the IO scheduling layer.
The request processing method according to claim 2, wherein the IO scheduling layer includes an IO scheduling interface, and the IO scheduling interface accesses different scheduling policies in the form of a plug-in.
The request processing method according to claim 3, wherein the parsing of the received SCSI request by using a scheduling strategy comprises:

The scheduling strategy is used to analyze the received SCSI requests in the current time period.
The request processing method according to claim 1, wherein if the size of the data to be processed corresponding to the SCSI request exceeds a predetermined threshold, the request processing method further comprises:

Judging whether the to-be-processed data corresponding to the SCSI request is multi-copy data;

If so, the SCSI request is split into multiple SCSI sub-requests according to the number of copies of the data to be processed, and each SCSI sub-request is encapsulated and sent to the software stack, so that the software stack passes the encapsulated request through the message The routing system sends to the server node corresponding to each SCSI sub-request.
The request processing method according to any one of claims 1 to 5, wherein the use of a scheduling strategy to analyze received SCSI requests and perform request merging processing includes:

If there are at least two SCSI requests that are connected in the first digit of the address, merge the SCSI requests connected in the first digit of the address into one request;

If the storage objects corresponding to at least two SCSI requests are the same storage object, the SCSI requests of the same storage object are combined into one request;

If the server nodes corresponding to at least two SCSI requests are the same server node, the SCSI requests of the same server node are combined into one request.
A request processing device based on a distributed storage system, characterized in that, based on an IO scheduling layer, the request processing device includes:

The receiving module is used to receive the SCSI request sent by the application layer; the analyzing module is used to analyze the received SCSI request using the scheduling strategy when the size of the data to be processed corresponding to the SCSI request does not exceed a predetermined threshold, and execute the request Merge processing

Encapsulation module, used to encapsulate the processed request separately;

The sending module is used to send the encapsulated request to the software stack, so that the software stack sends the encapsulated request to the corresponding server node through the message routing system.
The request processing device according to claim 7, further comprising:

A judging module, configured to judge whether the to-be-processed data corresponding to the SCSI request is multi-copy data when the size of the to-be-processed data corresponding to the SCSI request exceeds a predetermined threshold;

A splitting module, configured to split the SCSI request into multiple SCSI sub-requests according to the number of copies of the to-be-processed data when the to-be-processed data corresponding to the SCSI request is multiple copies of data;

The encapsulation module is used to encapsulate each SCSI sub-request;

The sending module is used to send each encapsulated SCSI sub-request to the software stack, so that the software stack sends the encapsulated request to the server node corresponding to each SCSI sub-request through the message routing system.
A request processing device based on a distributed storage system, which is characterized in that it comprises:

Memory, used to store computer programs;

The processor is configured to implement the steps of the request processing method according to any one of claims 1 to 6 when the computer program is executed.
A computer-readable storage medium, wherein a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the request processing method according to any one of claims 1 to 6 is realized A step of.