US-12627734-B2 - Sharing system based on multiple device nodes and method and storage medium

US12627734B2US 12627734 B2US12627734 B2US 12627734B2US-12627734-B2

Abstract

Disclosed is a sharing system based on multiple device nodes. The sharing system comprises: a host and a storage system, and host and each device node each are configured with a shared network card; where each device node comprises: a storage network device, where storage network device is configured to control a local device node to broadcast interconnection information to another device node in storage system, and establish, based on response information received by local device node, preset connection between local device node and a target device node; a lossless network device, where lossless network device is configured to manage a network environment of storage system; and a management device, where management device is configured to synchronize node data of local device node from shared network card to a shared network card of target device node, and manage preset connection between local device node and target device node.

Inventors

Shihui Zhang
Shuai ZHAO
Qinglin Liu

Assignees

SUZHOU METABRAIN INTELLIGENT TECHNOLOGY CO., LTD.

Dates

Publication Date: 20260512
Application Date: 20230619
Priority Date: 20221130

Claims (20)

1 . A sharing system based on multiple device nodes, comprising: a host, wherein the host is configured with a shared network card; and a storage system, wherein the storage system comprises the multiple device nodes, the multiple device nodes are connected to the host through a router, and each of the multiple device nodes is configured with a shared network card; wherein each of the multiple device nodes comprises: a storage network device, wherein the storage network device is configured to control a local device node to broadcast interconnection information to another device node in the storage system, and establish, based on response information received by the local device node, a preset connection between the local device node and a target device node indicated by a node identifier carried in the response information; a lossless network device, wherein the lossless network device is configured to manage a network environment of the storage system; and a management device, wherein the management device is configured to synchronize node data of the local device node from the shared network card to a shared network card of the target device node, and manage the preset connection between the local device node and the target device node.
2 . The sharing system according to claim 1 , wherein the device node further comprises: an environment deployment device, wherein the environment deployment device is configured to deploy the network environment after the device node in the storage system is activated; and/or, a memory management device, wherein the memory management device is configured to apply for a memory from an operating system of the host, and manage apparatus memories of each apparatus in the storage network device, the lossless network device, and the management device that share the shared network card.
3 . The sharing system according to claim 1 , wherein the storage network device comprises: a port address management apparatus, wherein the port address management apparatus is configured to manage an address of a preset port of the local device node, and monitor a port number of the preset port, and the preset port is a port for data transmission between the local device node and the another device node; and a port information management apparatus, wherein the port information management apparatus is configured to synchronize information of the preset port of the local device node to the another device node, and register protocol information of a communication protocol at the local device node.
4 . The sharing system according to claim 1 , wherein the storage network device further comprises: a broadcast apparatus, wherein the broadcast apparatus is configured to broadcast the interconnection information to the another device node at a preset interval period and receive the response information broadcast by the another device node; the interconnection information and the response information comprise at least one of the following: port information, protocol information, address information, and network information; and a type of broadcasting the interconnection information is divided into first multicast and second multicast, wherein the first multicast is configured to discover the another device node from a first-type network, and the second multicast is configured to discover the another device node from a second-type network.
5 . The sharing system according to claim 4 , wherein the storage network device further comprises: a connection management apparatus, wherein the connection management apparatus is configured to identify a to-be-connected device node based on the response information received by the broadcast apparatus, and establish the preset connection between the local device node and the to-be-connected device node; and break the preset connection between the local device node and the target device node in a case that the target device node having a preset connection relation with the local device node is offline.
6 . The sharing system according to claim 5 , wherein the connection management apparatus initiates the preset connection through a manner as follows: receiving, through a discovery process, the response information notified by a daemon process, and notifying the router that the local device node is in an online state; receiving, through the discovery process, information that the another device node is online, and notifying the daemon process that the another device node is online; and determining, through the daemon process, whether the another device node is configured with a discovery sub-net, determining whether an address of the another device node belongs to the discovery sub-net in a case that the another device node is configured with the discovery sub-net, querying the response information of the another device node in a case that the address of the another device node belongs to the discovery sub-net, and initiating the preset connection to the another device node based on the response information queried.
7 . The sharing system according to claim 1 , wherein the storage network device further comprises: a multi-queue management apparatus, wherein the multi-queue management apparatus is configured to remotely manage a memory data access queue of the local device node, and provide a data transmission interface for a dynamic link management apparatus of the management device; and the dynamic link management apparatus is configured to manage the preset connection between the local device node and the another device node and a connection between the local device node and the host.
8 . The sharing system according to claim 1 , wherein the storage network device further comprises: a communication protocol management apparatus, wherein the communication protocol management apparatus is configured to manage a communication protocol of the local device node, transmit a protocol instruction of the communication protocol, receive a protocol state of the communication protocol, and perform data transmission and protocol processing based on the protocol instruction.
9 . The sharing system according to claim 1 , wherein the lossless network device comprises: a data packet length adjustment apparatus, wherein the data packet length adjustment apparatus is configured to adjust a maximum number of bytes of a data packet transmitted through the preset connection associated with the local device node.
10 . The sharing system according to claim 1 , wherein the lossless network device further comprises: a flow isolation apparatus, wherein the flow isolation apparatus is configured to divide a preset port of the local device node into different regions, and perform flow isolation through the different regions.
11 . The sharing system according to claim 1 , wherein the lossless network device further comprises: a flow control apparatus, wherein the flow control apparatus is configured to update flow configuration information of a preset port of the local device node; and a congestion control apparatus, wherein the congestion control apparatus is configured to update congestion configuration information of the preset port of the local device node.
12 . The sharing system according to claim 1 , wherein the management device comprises: a port information identification apparatus, wherein the port information identification apparatus is configured to describe port information of the local device node, identify a preset port of the another device node, and determine whether the preset port of the another device node belongs to the same device node.
13 . The sharing system according to claim 1 , wherein the management device further comprises: a node information synchronization apparatus, wherein the node information synchronization apparatus is configured to synchronize the node data to a cluster control apparatus of the local device node and a cluster control apparatus of the target device node; and the node data comprise address information, port information, protocol information, and configuration information.
14 . The sharing system according to claim 13 , wherein the management device further comprises: an event management apparatus, wherein the event management apparatus is configured to synchronize data between the node information synchronization apparatus and the cluster control apparatus; synchronize data of the cluster control apparatus of the local device node to the cluster control apparatus of the target device node; and receive data synchronized by an event management apparatus of the target device node, and notify the cluster control apparatus of the local device node and the node information synchronization apparatus.
15 . The sharing system according to claim 14 , wherein the management device further comprises: a dynamic link management apparatus, wherein the dynamic link management apparatus is configured to manage the preset connection between the local device node and the another device node and a connection between the local device node and the host; wherein handover to a backup link is performed in a case that any one of the preset connections is broken, the node data are synchronized to a peer device node of the local device node in a case that there is no backup link, the cluster control apparatus is notified of information that there is no backup link through the peer device node, and the information that there is no backup link is synchronized to the event management apparatus through the cluster control apparatus; and host information is acquired from the node data in the cluster control apparatus in a case that the event management apparatus receives the information that there is no backup link, and connection to a host indicated by the host information is initiated.
16 . A sharing method based on multiple device nodes, applied to a sharing system, wherein the sharing system comprises: a host, wherein the host is configured with a shared network card; and a storage system, wherein the storage system comprises multiple device nodes, the multiple device nodes are connected to the host through a router, and each of the multiple device nodes is configured with a shared network card; each of the multiple device nodes comprises: a storage network device, a lossless network device, and a management device, wherein the lossless network device is configured to manage a network environment of the storage system; and the method comprises: controlling, by the storage network device, a local device node to broadcast interconnection information to another device node in the storage system, and establishing, based on response information received by the local device node, a preset connection between the local device node and a target device node indicated by a node identifier carried in the response information; and synchronizing, by the management device, node data of the local device node from the shared network card to a shared network card of the target device node, and managing the preset connection between the local device node and the target device node.
17 . The method according to claim 16 , wherein the controlling, by the storage network device, a local device node to broadcast interconnection information to another device node in the storage system, and establishing, based on response information received by the local device node, a preset connection between the local device node and a target device node indicated by a node identifier carried in the response information comprises: broadcasting, at a preset interval period by a broadcast apparatus in the storage network device, the interconnection information to the another device node and receiving the response information broadcast by the another device node, wherein the interconnection information and the response information comprise at least one of the following: port information, protocol information, address information, and network information; and a type of broadcasting the interconnection information is divided into first multicast and second multicast, wherein the first multicast is configured to discover the another device node from a first-type network, and the second multicast is configured to discover the another device node from a second-type network; and identifying, by a connection management apparatus in the storage network device, a to-be-connected device node based on the response information received by the broadcast apparatus, and establishing a preset connection between the local device node and the to-be-connected device node.
18 . The method according to claim 16 , wherein the establishing, based on response information received by the local device node, a preset connection between the local device node and a target device node indicated by a node identifier carried in the response information comprises: receiving, by a connection management apparatus in the storage network device through a discovery process, the response information notified by a daemon process, and notifying the router that the local device node is in an online state; receiving, by the connection management apparatus through the discovery process, online information of the another device node, and notifying the daemon process that the another device node is online; determining, by the connection management apparatus through the daemon process, whether the another device node is configured with a discovery sub-net, and determining whether an address of the another device node belongs to the discovery sub-net in a case that the another device node is configured with the discovery sub-net; and querying, by the connection management apparatus, the response information of the another device node in a case that the address of the another device node belongs to the discovery sub-net, and initiating the preset connection to the another device node based on the response information queried.
19 . The method according to claim 16 , wherein the managing the preset connection between the local device node and the target device node comprises: performing handover to a backup link in a case that any one of the preset connections is broken, synchronizing the node data to a peer device node of the local device node in a case that there is no backup link, notifying a cluster control apparatus of information that there is no backup link through the peer device node, and synchronizing the information that there is no backup link to an event management apparatus through the cluster control apparatus; and acquiring host information from the node data in the cluster control apparatus in a case that the event management apparatus receives the information that there is no backup link, and initiating connection to a host indicated by the host information.
20 . A non-volatile readable storage medium, storing a computer program, wherein when executed by a processor, the computer program implements steps of the sharing method based on multiple device nodes according to claim 16 .

Description

CROSS-REFERENCE TO RELATED APPLICATION The present application is a National Stage Application of PCT International Application No. PCT/CN2023/101165 filed on Jun. 19, 2023, which claims priority to Chinese Patent Application 202211533650.3, filed in the China National Intellectual Property Administration on Nov. 30, 2022, the disclosure of which is incorporated herein by reference in its entirety. TECHNICAL FIELD The disclosure relates to the technical field of data processing, and in particular to a sharing system based on multiple device nodes, a sharing method based on multiple device nodes and a non-volatile readable storage medium. BACKGROUND With the development of Internet, cloud computation, Internet of things, mobile Internet, etc., data grow explosively. Moreover, new challenges are presented to conventional data transmission networks. In the face of high-performance computation, big data analysis, and surge type input/output (IO) applications with high concurrency and low delay, a current transmission control protocol/Internet protocol (TCP/IP) hardware and software architecture and a technical feature of applying high central processing unit (CPU) consumption cannot satisfy demands of applications at all. In the related art, remote direct memory access (RDMA) can directly transfer data from a memory of one computer to another computer in a messaging passing through kernel (MPTK) manner, without intervention of both operating systems. Therefore, low latency, low CPU overhead, and a high bandwidth are realized. To be compatible with an existing Ethernet network, RDMA over Ethernet (ROCE) is developed on top of the RDMA, which wins an RDMA technology substantial development. However, owing to the unreliable Ethernet, it is impossible for device nodes on each network to perceive network transmission clogging or congestion, and network handover. In consequence, new challenges are presented to a storage system having high requirements on delay, bandwidth, and fault management. No effective solution to the above problems has been provided yet. SUMMARY A sharing system based on multiple device nodes is provided in an example of the disclosure, so as to at least solve the technical problem that in the related art, a storage system cannot enable multiple device nodes thereof to share a remote direct memory access (RDMA) over an Ethernet (ROCE) network card with high performance. In an aspect, a sharing system based on multiple device nodes is provided according to an example of the disclosure. The sharing system comprises: a host, wherein the host is configured with a shared network card; and a storage system, wherein the storage system comprises the multiple device nodes, the multiple device nodes are connected to the host through a router, and each of the multiple device nodes is configured with a shared network card; wherein each of the multiple device nodes comprises: a storage network device, wherein the storage network device is configured to control a local device node to broadcast interconnection information to another device node in the storage system, and establish, based on response information received by the local device node, a preset connection between the local device node and a target device node indicated by a node identifier carried in the response information; a lossless network device, wherein the lossless network device is configured to manage a network environment of the storage system; and a management device, wherein the management device is configured to synchronize node data of the local device node from the shared network card to a shared network card of the target device node, and manage the preset connection between the local device node and the target device node. Optionally, the device node further comprises: an environment deployment device, wherein the environment deployment device is configured to deploy the network environment after the device node in the storage system is activated. Optionally, the device node further comprises: a memory management device, wherein the memory management device is configured to apply for a memory from an operating system of the host, and manage apparatus memories of each apparatuses in the storage network device, the lossless network device, and the management device that share the shared network card. Optionally, the storage network device comprises: a port address management apparatus, wherein the port address management apparatus is configured to manage an address of a preset port of the local device node, and monitor a port number of the preset port, and the preset port is a port for data transmission between the local device node and the another device node; and a port information management apparatus, wherein the port information management apparatus is configured to synchronize information of the preset port of the local device node to the another device node, and register protocol information of a communication protocol at the l