WO2022253087A1

WO2022253087A1 - Data transmission method, node, network manager, and system

Info

Publication number: WO2022253087A1
Application number: PCT/CN2022/095142
Authority: WO
Inventors: 周超; 徐世萍
Original assignee: 华为技术有限公司
Priority date: 2021-05-31
Filing date: 2022-05-26
Publication date: 2022-12-08
Also published as: CN113472646A; CN113472646B

Abstract

A data transmission method, a node, a network manager, and a system, for use in realizing quick repair of a transmission path fault. The method comprises: the network manager determines a main transmission path and a standby transmission path between a source node and a destination node according to a transmission path request from the source node which comprises an identifier of the destination node, and delivers the main transmission path and the standby transmission path to the source node; the source node transmits a data message by using the standby transmission path when the main transmission path fails to transmit the data message.

Description

A data transmission method, node, network manager and system

Cross References to Related Applications

This application claims the priority of the Chinese patent application submitted to the China Patent Office on May 31, 2021, with the application number 202110604511.4 and the application name "A data transmission method, node, network manager and system", the entire contents of which are passed References are incorporated in this application.

technical field

The present application relates to the technical field of communications, and in particular to a data transmission method, node, network manager and system.

Background technique

In enterprise data centers or supercomputing centers, the network architecture used by computing clusters for high-performance computing (high performance computing, HPC) services is usually an infinite bandwidth (infiniband, IB) network or remote direct memory access (DRAM) based on converged Ethernet. RDMA over converged ethernet (RoCE) network, in which RoCE network allows the use of remote direct memory access (remote direct memory access, RDMA) technology in traditional Ethernet, based on the mature ecological implementation of traditional Ethernet, the cost has obvious advantages compared with IB network, And the version evolution rate is much faster than other types of networks, and will become the mainstream network selection for HPC services in the future.

At present, in order to optimize the forwarding performance of the RoCE network, the source routing forwarding mechanism is usually used to replace the traditional Internet protocol (internet protocol, IP) routing forwarding mechanism. As shown in Figure 1, before the data packet is transmitted, the network controller first issues the source routing path of the transmission path to the source node, and the source routing path includes each node on the transmission path where the source node transmits the data packet to the destination node The identifier of the outgoing port of each node in the network, and the outgoing port of each node in the network is directly connected to another node in the network. By carrying the information of the source routing path in the data message, each node on the transmission path can forward the data Message, to realize the transmission of data message from the source node to the destination node. However, in the existing source routing mechanism, the source node and the destination node rely on periodic heartbeat connections to detect whether there is a link failure between the source node and the destination node, and re-request the network controller to configure the source node and the destination node after the failure is detected. For source routing between destination nodes, the repair time for faults is longer, and the user experience is poor.

Contents of the invention

The present application provides a data transmission method, a node, a network manager and a system, which are used to realize rapid recovery of transmission path failures.

In a first aspect, the present application provides a data transmission system, the data transmission system can perform data transmission based on a RoCE network, including a network manager, a source node and a destination node; the source node is used to send the network manager A transmission path request, where the transmission path request includes the identifier of the destination node; the network manager is configured to determine a main transmission path and a backup path between the source node and the destination node according to the transmission path request a transmission path, before the source node transmits a data message to the destination node, sending a transmission path response including the primary transmission path and the standby transmission path to the source node; the source node is also used to record the The main transmission path and the transmission path of the backup transmission path respond, and when the failure occurs in the transmission of the data packet by the primary transmission path, the backup transmission path is used to transmit the data packet.

Using the above method, after receiving the transmission path request from the source node, the network manager not only determines the main transmission path for the data message transmission of the source node, but also determines one or more backup transmission paths for the data message transmission of the source node , and return the main transmission path and the backup transmission path to the source node. When the main transmission path fails, the source node can use the backup transmission path to transmit data packets, and there is no need to request the network manager to reconfigure the transmission path, which is conducive to the realization of transmission Quick repair of path faults improves user experience.

In a possible design, there is no intersection between the main transmission path and the backup transmission path as much as possible, that is, the main transmission path and the backup transmission path share the least number of nodes between the source node and the destination node.

In the above design, it is beneficial to avoid the situation that both the primary transmission path and the backup transmission path cannot transmit data packets due to failure of the same node.

In a possible design, the data transmission system further includes: a first intermediate node, wherein the first intermediate node is any one located between the source node and the destination node in the main transmission path node; the first intermediate node is configured to determine a fallback path according to the data message; the first intermediate node is also configured to determine the fallback path according to the fallback path Send a fault notification message to the source node.

In the above design, when the first intermediate node detects the failure of the main transmission path of the transmission data message, it sends a failure notification message to the source node, which is conducive to the rapid perception of the failure of the main transmission path by the source node and realizes the rapid detection of the failure of the transmission path. repair.

In another possible design, the data transmission system further includes: one or more second intermediate nodes, wherein the one or more second intermediate nodes are located between the source node and the main transmission path A node between the first intermediate nodes; when the first intermediate node determines a fallback path according to the data message, it is specifically used to: according to the data message carried in the data message, in the One or more identifiers of ingress ports of the second intermediate node, to determine the fallback path.

In the above design, by carrying the identification of the ingress port of the data message in the second intermediate node in the data message, it is beneficial for the first intermediate node to quickly trace back the source of the fault notification message.

In another possible design, the first intermediate node is further configured to, before forwarding the data message to the next-hop node adjacent to the first intermediate node in the main transmission path, An identifier of an ingress port of the data message at the first intermediate node is added to the data message.

In the above design, when the first intermediate node forwards the data message, it adds the identification of the ingress port of the data message in the data message to the data message, which is beneficial to the subsequent nodes of the first intermediate node when they find that the transmission path is faulty. , to trace back the source of the fault notification message.

In another possible design, when the first intermediate node adds the identifier of the ingress port of the data packet in the data packet to the data packet, it is specifically used to send the data packet The identifier of the egress port of the first intermediate node recorded in the source routing label carried in the file is replaced by the identifier of the ingress port.

In the above design, the outbound port field in the source routing label carried by the multiplexed data message carries the inbound port identifier, which is beneficial to avoid modification of the data message transmission format and improve the forwarding efficiency of the data message.

In a second aspect, the present application provides a data transmission method, which includes: a source node sends a transmission path request to a network manager, and the transmission path request includes an identifier of a destination node; The transmission path response of the device including the main transmission path and the backup transmission path, and record the transmission path response including the main transmission path and the backup transmission path; when the main transmission path fails to transmit data packets, the The source node transmits the data packet by using the standby transmission path.

In a possible design, the method further includes: when the source node receives a failure notification message from the first intermediate node, determining that a failure occurs in the data message transmitted by the main transmission path, wherein the first An intermediate node is any node located between the source node and the destination node in the main transmission path.

In a third aspect, the present application provides a data transmission method, the method comprising: a network manager receives a transmission path request from a source node, and the transmission path request includes an identifier of a destination node; A main transmission path and a standby transmission path are determined between the node and the destination node; the network manager sends a message including the main transmission path and the backup transmission path to the source node before the source node transmits a data message to the destination node. The transport path response for the transport path.

In a fourth aspect, the present application provides a data transmission method, the method comprising: a first intermediate node receiving a data message from a source node, wherein the first intermediate node is located between the source node and the data message The destination node transmits the node in the main transmission path of the data message; the first intermediate node determines the fallback path according to the data message; when detecting that the main transmission path fails, the first intermediate node The intermediate node sends a fault notification message to the source node according to the fallback path.

In a possible design, the first intermediate node determines a fallback path according to the data message, including: the first intermediate node determines the fallback path according to the data message carried in the data message in a or a plurality of identifications of ingress ports of second intermediate nodes, and determine the fallback path, wherein the one or more second intermediate nodes are located between the source node and the first intermediate node in the main transmission path nodes between.

In another possible design, the method further includes: before forwarding the data packet to the next-hop node adjacent to the first intermediate node in the main transmission path, the first intermediate node The node adds the identifier of the ingress port of the data packet at the first intermediate node to the data packet.

In another possible design, the first intermediate node adds an identifier of the ingress port of the data message in the data message to the data message, including: the first intermediate node will The identifier of the egress port of the first intermediate node recorded in the source routing label carried in the data packet is replaced with the identifier of the ingress port.

In the fifth aspect, the embodiment of the present application provides a data transmission device, which has the function of implementing each step in the above-mentioned second aspect and any possible design of the second aspect. The function can be realized by hardware, or by The hardware executes the corresponding software implementation. The hardware or software includes one or more units (modules) corresponding to the above functions, such as a communication unit and a processing unit.

In one possible design, the device may be a chip or an integrated circuit.

In a possible design, the device includes a processor and an interface circuit, the processor is coupled to the interface circuit, and is used to realize the functions of each step in the second aspect and any possible design of the second aspect . It can be understood that the interface circuit may be a transceiver or an input/output interface. The device may further include a memory, where the memory stores a program executable by the processor for realizing the functions of each step in the above-mentioned second aspect and any possible design of the second aspect.

In one possible design, the device may be a source node.

In the sixth aspect, the embodiment of the present application provides a data transmission device, which has the function of realizing each step in the third aspect above, and the function can be realized by hardware, or by executing corresponding software by hardware. The hardware or software includes one or more units (modules) corresponding to the above functions, such as a communication unit and a processing unit.

In one possible design, the device may be a chip or an integrated circuit.

In a possible design, the device includes a processor and an interface circuit, the processor is coupled to the interface circuit, and is configured to implement the functions of the steps in the above third aspect. It can be understood that the interface circuit may be a transceiver or an input-output interface. The device may further include a memory storing a program executable by the processor for realizing the functions of each step in the above third aspect.

In one possible design, the device may be a network manager.

In the seventh aspect, the embodiment of the present application provides a data transmission device, which has the function of implementing each step in the fourth aspect and any possible design of the fourth aspect. The function can be realized by hardware, or by The hardware executes the corresponding software implementation. The hardware or software includes one or more units (modules) corresponding to the above functions, such as a communication unit and a processing unit.

In one possible design, the device may be a chip or an integrated circuit.

In a possible design, the device includes a processor and an interface circuit, the processor is coupled to the interface circuit, and is used to realize the functions of each step in the fourth aspect and any possible design of the fourth aspect . It can be understood that the interface circuit may be a transceiver or an input/output interface. The device may further include a memory, the memory stores a program executable by the processor for realizing the function of each step in the fourth aspect and any possible design of the fourth aspect.

In a possible design, the device may be a first intermediate node.

In the eighth aspect, the embodiment of the present application also provides a computer program, which, when the computer program is run on a computer, causes the computer to execute any one of the above-mentioned second to fourth aspects and the second to fourth aspects Methods provided in Possible Designs.

In the ninth aspect, the embodiment of the present application also provides a computer-readable storage medium, where a computer program is stored in the computer-readable storage medium, and when the computer program is executed by a computer, the computer executes the above-mentioned second to the fourth aspect and the method provided in any possible design of the second to fourth aspects.

In the tenth aspect, the embodiment of the present application also provides a chip, the chip is used to read the computer program stored in the memory, and execute any one of the possible tasks in the second to fourth aspects and the second to fourth aspects above. method provided in the design.

For the technical effects that can be achieved from the second aspect to the tenth aspect, please refer to the technical effects that can be achieved by the first aspect above, and will not be repeated here.

Description of drawings

FIG. 1 is a schematic diagram of a source routing forwarding mechanism;

FIG. 2 is a schematic diagram of the principle of an ECMP mechanism provided in the embodiment of the present application;

FIG. 3 is a schematic diagram of the principle of an AR mechanism provided by an embodiment of the present application;

FIG. 4 is one of the schematic diagrams of the network architecture provided by the embodiment of the present application;

FIG. 5 is one of the schematic diagrams of the architecture of the data transmission system provided by the embodiment of the present application;

FIG. 6 is a schematic flowchart of a data transmission method provided by an embodiment of the present application;

FIG. 7 is the second schematic diagram of the network architecture provided by the embodiment of the present application;

FIG. 8 is one of schematic diagrams of a source routing label in an embodiment of the present application;

FIG. 9 is a second schematic diagram of a source routing label according to an embodiment of the present application;

FIG. 10 is one of the schematic diagrams of the data transmission device provided by the embodiment of the present application;

Figure 11 is the second schematic diagram of the data transmission device provided by the embodiment of the present application;

FIG. 12 is the second schematic diagram of the architecture of the data transmission system provided by the embodiment of the present application.

Detailed ways

Before introducing the embodiments of the present application, some terms in the present application are firstly explained, so as to facilitate the understanding of those skilled in the art.

1) Host, also known as service node, communication node, etc., can be used to process business data, has the function of sending and receiving, and can send data to other hosts and/or receive data from other hosts. For example: the host can be a server, a big data platform, the cloud, a cloud server, a server cluster, a terminal device or a computer device, etc., or it can be a component of these devices, such as a chip or a chip system. Among them, computer equipment, which may be referred to as a computer for short, refers to equipment with processing functions such as network data storage, data transmission, and data reception.

2), transmission node, also known as intermediate node, switching node, etc., is a device with data exchange (forwarding) function, which can be a switch, router, gateway and other devices, or other devices with data exchange A functional device or device, or a component of these devices, such as a chip or a chip system. This embodiment of the present application does not limit it.

3) The IP routing and forwarding mechanism usually refers to the mechanism based on message information (such as the destination IP address, etc.) using a routing selection algorithm (such as a hash algorithm, etc.) to implement routing and forwarding. Take the equal cost multipath (equal cost multipath, ECMP) mechanism and adaptive routing (adaptive routing, AR) in the IP routing and forwarding mechanism as examples. As shown in Figure 2, the ECMP mechanism uses the hash method to calculate the outgoing ports of different data streams forwarded based on the five-tuple (source IP address, source port, destination IP address, destination port, and transport layer protocol), and completes each data stream and One-to-one mapping of end-to-end transmission paths, different data streams are evenly distributed to each end-to-end transmission path. Since the quintuple of each flow is determined, the egress port of each ECMP hash is also uniquely determined, and the end-to-end transmission path of the flow is finally uniquely determined. However, the biggest problem with the ECMP load sharing scheme is that when the traffic in the network is unevenly distributed (mixed elephant and mouse flows), treating large flows and small flows as equivalent and assigning them to different transmission paths will cause each transmission The load is heavily unbalanced among the paths.

As shown in Figure 3, the AR mechanism refers to the ECMP mechanism. On the basis of the ECMP mechanism, the judgment of the congestion status of the outgoing port queue is added. If the congestion exceeds the threshold, it is adjusted to other ports. The load is severely unbalanced. However, both the ECMP mechanism and the AR mechanism require each intermediate node on the transmission path to maintain the global routing table of the entire network and execute the routing algorithm, which is difficult to maintain and implement, and the transmission delay is relatively large because each intermediate node needs to execute the routing algorithm. , not as simple in concept as the source routing mechanism, and easy to implement.

4) Source routing, which can also be called a source routing mechanism, that is, the source node of data transmission (ie, the sending node) can specify some or all of the transmission nodes (or intermediate nodes) that the sent message passes along the way. As shown in Figure 1, it is known that the outgoing port A of the source node is connected to the intermediate node 1, the outgoing port B of the intermediate node 1 is connected to the intermediate node 2, the outgoing port C of the intermediate node 2 is connected to the intermediate node 3, and the outgoing port B of the intermediate node 3 is connected to the intermediate node 3. The outgoing port D is connected to the intermediate node 4, the outgoing port E of the intermediate node 4 is connected to the intermediate node 5, and the outgoing port F of the intermediate node 5 is connected to the destination node. Port A, the outgoing port B of the intermediate node 1, the outgoing port C of the intermediate node 2, the outgoing port D of the intermediate node 3, the outgoing port E of the intermediate node 4, and the outgoing port F of the intermediate node 5 (that is, the identification of the transmission path Source routing path) indicates that each hop node that the data packet passes through from the source node is intermediate node 1, intermediate node 2, intermediate node 3, intermediate node 4, intermediate node 5 and the destination node.

However, in the existing source routing mechanism, the source node can only rely on the periodic heartbeat connection with the destination node to detect whether there is a link failure between the source node and the destination node, and re-request the network controller to send the source node after the failure is detected. For the source routing path of the new transmission path between the node and the destination node, the efficiency of fault detection is low, and after the fault is found, the network controller needs to re-scan the network architecture to configure the new transmission path, which takes a long time. In this case, It is especially serious when the network architecture is complex, and it is difficult to quickly repair transmission path failures.

In view of this, the embodiment of the present application provides a data transmission solution, which is used to determine the main transmission path and the backup transmission path at the source node and the destination node through the network manager at the same time, so as to realize the rapid recovery after the failure of the transmission path. Embodiments of the present application will be described in detail below in conjunction with the accompanying drawings.

As shown in Figure 4, it is an example of a data transmission system under a possible fat-tree (Fat-Tree) network architecture applicable to the embodiment of the present application, including multiple hosts (such as host H11, host H12, etc.), multiple intermediate Nodes (such as intermediate node E11, intermediate node A11, intermediate node C11, etc.) and a network manager (fabric manager, FM). Among them, the intermediate nodes can also be called transmission nodes, and the intermediate nodes of the leaf (leaf) layer and the spine (spine) layer can be divided into different delivery units (point of delivery, POD) or clusters, and each leaf in each POD The intermediate nodes of each layer are connected to the intermediate nodes of each spine layer; at the same time, the intermediate nodes of each spine layer can be connected to the intermediate nodes of one or more super-spine layers, so that different PODs can pass through the super-spine The intermediate nodes of the (super-spine) layer are connected. Among them, the network manager has a network management function, is directly or indirectly connected to each intermediate node and host computer in the network, and can provide network transmission path configuration. During data transmission, any host in FIG. 4 can be used as a source node of data transmission, that is, a data sending end, or as a destination node of data transmission, that is, a data receiving end.

In order to realize the data transmission scheme of the present application, the host (such as host H11, host H12 etc.) among Fig. , intermediate node C11, etc.) may include components such as a switch (Switch) component and a switch agent (Switch-agent) component, and the network manager may include a global network manager controller (FM-controller) component, a global network manager management (FM-manager) components and other components. Specifically, as shown in Figure 5, the components of the network manager (FM), host (Host) and intermediate nodes (such as switches (Switch)) in the data transmission system can be divided into forwarding plane, control plane and The management plane has three levels, where the forwarding plane can include the RNIC network card on the host side and the Switch component on the intermediate node side; the control plane includes the Host-agent component on the host side, the Switch-agent component on the intermediate node side, and the FM on the network manager side -controller component; the management plane includes the FM-manager component on the network manager side. The FM-manager component can deliver port configurations (port config) to intermediate nodes, and can also learn the port link status (port link status) and port input/output adaptation rate (port input/output utility rate) between intermediate nodes, etc. information, and can store network topology (network topology) and other information in a database (data base, DB). The data transmission system can perform data transmission based on the RoCE network. When transmitting data packets, the RNIC network card on the host side can request a transmission path from the FM-controller component, and the FM-controller component returns the main transmission path to the RNIC network card on the host side. path and backup transmission path, and then the RNIC network card on the host side can transmit data packets through the main transmission path and backup transmission path and other RNIC network cards on the host side.

FIG. 6 is a schematic diagram of a data transmission method provided in an embodiment of the present application, the method including:

S601: The source node sends a transmission path request to a network manager, and the network manager receives the transmission path request.

Wherein, the transmission path request includes the identifier of the destination node.

In some implementations, when an application (such as an HPC application) in the source node (such as the host H11) is started, it will trigger the source node to initiate a network connection (such as an RDMA connection) to the destination node (such as the host H21), for the source The data message of a certain service is transmitted between the node and the destination node. Specifically, when an application in the source node is started and initiates a network connection to the destination node, the source node may send a transmission path request including the identifier of the destination node to the network manager, requesting the network manager to configure the connection to the destination node for the source node. The transmission path of the data message sent by the node.

As an example, when the HPC application in the source node is started, when an RDMA connection is initiated to the destination node, the RDMA network interface card (RDMA network interface card, RNIC) of the source node may send the identification of the destination node to the network manager. The request for a transmission path, wherein the identifier of the destination node may be an identity document (ID), IP address, etc. of the destination node, which is not limited in this application.

S602: The network manager determines a primary transmission path and a standby transmission path between the source node and the destination node, and before the source node transmits a data packet to the destination node, sends a message including the Transmission path responses of the primary transmission path and the standby transmission path.

In the embodiment of this application, the network manager has a network management function, and is directly or indirectly connected to each intermediate node (also called a transmission node) and a host in the network, and can provide network transmission path configuration, performance control of each node, etc. Function. Between each node in the network (comprising intermediate node and host computer, and between intermediate node and intermediate node) can be connected by port, as shown in Figure 7, intermediate node E11 can be connected with the port 101 of intermediate node A11 by port 16. Specifically, the port connections between nodes in the network can be configured by network managers when nodes (including hosts and intermediate nodes) access the network, and the nodes report to the network manager, or can be configured by network managers through the network manager , and sent to corresponding nodes by the network manager, which is not limited in this application.

As an example, after receiving the transmission path request from the source node including the identification of the destination node, the network manager can determine the minimum transmission path between the source node and the destination node based on the topology of the managed network. path, that is, multiple transmission paths do not include repeated intermediate nodes as much as possible. As shown in Figure 7, the transmission path determined by the network manager includes transmission path 1: source node H11-intermediate node E11-intermediate node A11-intermediate node C11-intermediate node A21-intermediate node E21-destination node H21, and transmission path 2 : source node H11-intermediate node E11-intermediate node A12-intermediate node C22-intermediate node A22-intermediate node E21-destination node H21, wherein transmission path 1 and transmission path 2 are in the spine (spine) layer and super spine (super spine) Layers have no intersections, that is, no shared intermediate nodes.

After the network manager determines multiple transmission paths that share the least intermediate nodes between the source node and the destination node, one of the transmission paths can be determined as the main transmission path, and the other transmission paths are determined as backup transmission paths, and the transmission path including the Transmission path responses of the primary transmission path and the standby transmission path. Specifically, when the network manager determines the primary transmission path and the standby transmission path among the multiple transmission paths, it may use the transmission path with the least number of intermediate nodes as the primary transmission path according to the principle that the number of intermediate nodes is the least There are multiple transmission paths with the least number of nodes, and one of the transmission paths with the fewest number of intermediate nodes can be randomly selected as the main transmission path), and the other transmission paths can be used as backup transmission paths; of course, according to the principle of the smallest average load, The principle of the lowest transmission delay is to select the main transmission path among multiple transmission paths, and the other transmission paths are used as backup transmission paths.

Of course, the network manager may also directly send a transmission path response including the multiple transmission paths to the source node, and the source node selects one of the multiple transmission paths as the primary transmission path, and the other transmission paths as backup transmission paths.

Take source node H11-intermediate node E11-intermediate node A11-intermediate node C11-intermediate node A21-intermediate node E21-destination node H21 as the main transmission path, where source node H11 is connected to port 102 of intermediate node E11 through port 17 , the intermediate node E11 is connected to the port 103 of the intermediate node A11 through port 18, the intermediate node A11 is connected to the port 104 of the intermediate node C11 through the port 19, the intermediate node C11 is connected to the port 105 of the intermediate node A21 through the port 20, and the intermediate node A21 is connected through Port 21 is connected to port 106 of the intermediate node E21, and the intermediate node E21 is connected to port 107 of the destination node H21 through port 22, then the network manager can send the transmission path response of the source routing path including the main transmission path to the source node, Sending the main transmission path to the source node, wherein the source routing path of the main transmission path may include the identifiers of the egress ports of the nodes on the main transmission path that transmit data packets from the source node to the destination node. Wherein the identification of the outgoing port of each node may be composed of the identification of the node (such as an IP address) and the port number of the outgoing port. As an example, the source routing path of the main transmission path is shown in Table 1 below.

Table 1

Among them, the IP addresses of source node H11, intermediate node E11, intermediate node A11, intermediate node C11, intermediate node A21, intermediate node E21, and destination node H21 are IP H11, IP E11, IP A11, IP C11, IP A21, IP E21, IP H21.

S603: When the source node fails to transmit the data packet on the primary transmission path, the source node transmits the data packet by using the standby transmission path.

After the source node obtains the primary transmission path and the standby transmission path from the network manager, it records (or carries) the source routing path of the primary transmission path in the source routing label carried in the data message, and transmits to the destination node through the primary transmission path. datagram.

In a possible implementation, a heartbeat connection may also be maintained between the source node and the destination node to detect whether a failure occurs on the first transmission path. For example, the source node can send a heartbeat request message to the destination node through the main transmission path according to the set period (such as 1s, 2s), etc., if the heartbeat response message replied by the destination node is received within the specified time (such as within 1ms) , it means that the main transmission path can normally transmit packets without failure, otherwise, it is determined that the main transmission path is faulty.

In order to further improve the fault detection efficiency, in another possible implementation, after the data message is sent from the source node, each time the data message is forwarded through a hop, the intermediate node can mark the data message on the ingress port of the node Added to the data message to facilitate the traceability and rollback of the fault notification message when the transmission path fails.

As an example, when an intermediate node forwards a data packet, it may replace the identifier of the outgoing port of the intermediate node recorded in the source routing label with the identifier of the incoming port of the data packet at the intermediate node. Still taking the transmission of the data message through the above-mentioned main transmission path as an example, the data message carries the source routing label (that is, the message header) as shown in Figure 8, and the intermediate node E11 receives the data sent by the source node H11 through the port 102 (ingress port) message, when the intermediate node E11 forwards the data message through its own port 18 (outlet port) according to HOP1 (IP E11+18), HOP1 is changed from IP E11+18 to IP E11+102, similarly, the intermediate node C11 When forwarding data packets, modify HOP2 from IP A11+19 to IP A11+103, etc.

When any intermediate node in the main transmission path is detected to have a link failure on the outgoing port and the data message cannot be forwarded, the intermediate node will construct a failure notification (FN) message and forward it back from the incoming port of the data message. Return the fault notification message. Wherein the failure notification message includes the source routing path of the fallback path that sends the failure notification message to the source node (i.e. specifying the outgoing port of each hop on the fallback path), wherein the source routing path of the fallback path can be obtained from the datagram obtained from the text. For the convenience of subsequent descriptions, in the subsequent descriptions of the embodiments of the present application, the intermediate node on the main transmission path that detects the failure of the outgoing port link is referred to as the first intermediate node.

As an example, as shown in FIG. 7, the data message carrying the source routing label shown in FIG. After forwarding by the intermediate node A11 and the intermediate node C11 (multiple second intermediate nodes located between the source node and the first intermediate node), the intermediate node A21 (first intermediate node) on the main transmission path receives the data message. Wherein, when the intermediate node E11, the intermediate node A11 and the intermediate node C11 forward the data message, the identification of the outgoing port of the intermediate node recorded in the source routing label of the data message is replaced by the inbound port of the data message in the intermediate node. The identification of the port and the source route carried by the data packet received by the intermediate node A21 are shown in FIG. 9 . Assuming that the intermediate node A21 on the main transmission path detects that the port 21 forwarding the data message to the next hop node intermediate node E21 fails (such as the heartbeat connection interruption or electrical connection interruption between the port 21 and the intermediate node 21, etc.), the intermediate node A21 Determine the source routing path of the fallback path according to the source routing label in the data packet. As shown in Figure 9, the intermediate node A21 determines that the source routing path of the fallback path is Hop 0(IP A21+105), Hop1( IP C11+104), Hop2(IP A11+103), Hop3(IP E11+102).

The intermediate node A21 constructs and generates a fault notification message, and the source routing label carried by the fault notification message contains Hop 0 (IP A21+105), Hop1 (IP C11+104), Hop2 (IP A11+103), Hop3 (IP E11+102), the intermediate node A21 sends a fault notification message to the intermediate node C11 from the incoming port of the data message, namely port 105, according to the Hop1(IP C11+104) recorded in the source routing label carried by the fault notification message ), Hop2 (IP A11+103), Hop3 (IPE11+102), intermediate node C11, intermediate node A11, and intermediate node E11 forward the fault notification message hop by hop until the fault notification message is received by the host H11, and the transmission data is determined The main transmission path of the packet is faulty.

In addition, in order to improve the reliability of fault notification message transmission, intermediate nodes can reserve resources (such as reserved bandwidth resources, forwarding queue resources, etc.) Trace back.

In some implementations, the source node can transmit data packets of different services with multiple destination nodes at the same time. In order to facilitate the location of the faulty transmission path by the source node, the fault notification message can also carry the information of the service to which the corresponding data message belongs. Service flow identification, wherein the service flow identification of the service to which the data message belongs can be obtained from the flow label (flow label) field in the message header of the data message, etc., and the service flow identification can be the corresponding source IP, Destination IP + queue pair (queue pair, QP) identifier (identifier, ID), etc.

After the source node detects the failure of the main transmission path, it switches the transmission of data packets from the main transmission path to the standby transmission path issued by the network manager, that is, the source node uses the standby transmission path for data transmission, which can realize the fast transmission path. Failover to achieve the goal of fast convergence of business flows.

Specifically, after the source node detects that the primary transmission path fails, the source routing path in the source routing label carried by the data packet is switched from the source routing path of the primary transmission path to the source routing path of the standby transmission path, which can realize Fast failover of the transmission path achieves the goal of fast convergence of business flows.

In addition, it should be understood that the data transmission scheme provided by the embodiment of the present application is not only applicable to the Fat-Tree network architecture as shown in Figure 4, but also applicable to 3D/6D ring (torus) topology network architecture, dragonfly topological network architecture, etc.

The foregoing mainly introduces the solution provided by the present application from the perspective of interaction between the source node, the first intermediate node, and the network manager. It can be understood that, in order to realize the above functions, each network element (device) includes a corresponding hardware structure and/or software module (or unit) for performing each function. Those skilled in the art should easily realize that the present application can be implemented in the form of hardware or a combination of hardware and computer software in combination with the units and algorithm steps of each example described in the embodiments disclosed herein. Whether a certain function is executed by hardware or computer software drives hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art may use different methods to implement the described functions for each specific application, but such implementation should not be regarded as exceeding the scope of the present application.

FIG. 10 and FIG. 11 are schematic structural diagrams of a possible data transmission device provided by an embodiment of the present application. These data transmission devices can be used to implement the functions of the source node, the network manager, or the first intermediate node in the above method embodiments, and thus can also realize the beneficial effects of the above method embodiments. In the embodiment of the present application, the data transmission device may be the source node or the network manager or the first intermediate node, and may also be a module (such as a chip) applied to the source node or the network manager or the first intermediate node.

As shown in Figure 10. The data transmission device 1000 may include: a processing unit 1002 and a communication unit 1003 , and may also include a storage unit 1001 . The data transmission apparatus 1000 is configured to realize the functions of the source node or the network manager or the first intermediate node in the above method embodiments.

In a possible design, the processing unit 1002 is configured to implement corresponding processing functions. The communication unit 1003 is used to support communication between the data transmission device 1000 and other network entities. The storage unit 1001 is configured to store program codes and/or data of the data transmission device 1000 . Optionally, the communication unit 1003 may include a receiving unit and/or a sending unit, configured to perform receiving and sending operations respectively.

When the data transmission device 1000 is used to realize the function of the source node in the method embodiment: the communication unit 1003 is used to send a transmission path request to the network manager, and the transmission path request includes the identity of the destination node; The transmission path response of the network manager including the main transmission path and the backup transmission path;

The processing unit 1002 is configured to record a transmission path response including the primary transmission path and the backup transmission path;

The communication unit 1003 is further configured to use the standby transmission path to transmit the data packet when the primary transmission path fails to transmit the data packet.

In a possible design, the processing unit 1002 is further configured to, when the communication unit 1003 receives a failure notification message from the first intermediate node, determine that a failure occurs in the data message transmitted by the main transmission path, Wherein the first intermediate node is any node located between the source node and the destination node in the main transmission path.

When the data transmission device 1000 is used to realize the function of the network manager in the method embodiment: the communication unit 1003 is configured to receive a transmission path request from the source node, and the transmission path request includes the identifier of the destination node;

The processing unit 1002 is configured to determine a primary transmission path and a standby transmission path between the source node and the destination node;

The communication unit 1003 is further configured to send a transmission path response including the primary transmission path and the standby transmission path to the source node before the source node transmits the data packet to the destination node.

When the data transmission device 1000 is used to implement the function of the first intermediate node in the method embodiment: the communication unit 1003 is configured to receive a data message from the source node, wherein the first intermediate node is located at the source node A node in the main transmission path for transmitting the data message with the destination node of the data message;

The processing unit 1002 is configured to determine a fallback path according to the data message;

The communication unit 1003 is further configured to send a failure notification message to the source node according to the fallback path when a failure of the main transmission path is detected.

In a possible design, when the processing unit 1002 determines the fallback path according to the data message, it is specifically configured to: An identification of an ingress port of an intermediate node to determine the fallback path, wherein the one or more second intermediate nodes are nodes located between the source node and the first intermediate node in the main transmission path.

In a possible design, the processing unit 1002 is further configured to forward the data message to a next-hop node adjacent to the first intermediate node in the main transmission path in the communication unit 1003 Before, adding the identifier of the ingress port of the data packet at the first intermediate node to the data packet.

In a possible design, the processing unit 1002 is specifically configured to carry the The identifier of the egress port of the first intermediate node recorded in the source routing label of , is replaced by the identifier of the ingress port.

More detailed descriptions about the processing unit 1002 and the communication unit 1003 can be directly obtained by referring to related descriptions in the method embodiments, and details are not repeated here.

As shown in FIG. 11 , the data transmission device 1100 includes a processor 1110 and an interface circuit 1120 . The processor 1110 and the interface circuit 1120 are coupled to each other. It can be understood that the interface circuit 1120 may be an input and output interface. Optionally, the data transmission device 1100 may further include a memory 1130 for storing instructions executed by the processor 1110 or storing input data required by the processor 1110 to execute the instructions or storing data generated by the processor 1110 after executing the instructions.

When the data transmission device 1100 is used to implement the data transmission method applicable to the source node or the network manager or the first intermediate node, the processor 1110 is used to implement the functions of the above-mentioned processing unit 1002, and the interface circuit 1120 is used to implement the above-mentioned communication unit 1003 function.

As another form of this embodiment, a computer-readable storage medium is provided, on which instructions are stored, and when the instructions are executed, the above method can be executed data transfer method.

As another form of this embodiment, a computer program product including instructions is provided, and when the instructions are executed, the data transmission method applicable to the source node or the network manager or the first intermediate node in the above method embodiments can be executed.

As another form of this embodiment, a chip is provided, and when the chip is running, it can execute the data transmission method applicable to the source node or the network manager or the first intermediate node in the above method embodiments.

Fig. 12 is a schematic diagram of the architecture of a data transmission system provided by an embodiment of the present application. The data transmission system includes a network manager, a source node, a destination node, and a first intermediate node, wherein the network manager has the above-mentioned functions for realizing The function data transmission device of the network manager in the method embodiment, the source node has the above-mentioned function data transmission device for realizing the source node in the method embodiment, and the first intermediate node has the above-mentioned function data transmission device for realizing the first intermediate node in the method embodiment Function data transmission device.

The above-mentioned embodiments may be implemented in whole or in part by software, hardware, firmware or other arbitrary combinations. When implemented using software, the above-described embodiments may be implemented in whole or in part in the form of computer program products. The computer program product includes one or more computer instructions. When the computer program instructions are loaded or executed on the computer, the processes or functions according to the embodiments of the present application will be generated in whole or in part. The computer may be a general purpose computer, a special purpose computer, a computer network, or other programmable devices. The computer instructions may be stored in or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions may be transmitted from a website, computer, server or data center Transmission to another website site, computer, server, or data center by wired (eg, coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (eg, infrared, wireless, microwave, etc.). The computer-readable storage medium may be any available medium that can be accessed by a computer, or a data storage device such as a server or a data center that includes one or more sets of available media. The available media may be magnetic media (eg, floppy disk, hard disk, magnetic tape), optical media (eg, DVD), or semiconductor media. The semiconductor medium may be a solid state drive (SSD).

The foregoing is only a specific implementation manner of the present application. Those skilled in the art may conceive changes or substitutions based on the specific implementation methods provided in this application, and all of them shall fall within the protection scope of this application.

Claims

A data transmission system, characterized in that it includes a network manager, a source node and a destination node;

The source node is configured to send a transmission path request to the network manager, where the transmission path request includes the identifier of the destination node;

The network manager is configured to determine a primary transmission path and a backup transmission path between the source node and the destination node according to the transmission path request, and send a data packet to the destination node before the source node transmits a data message to the destination node. The source node sends a transmission path response including the primary transmission path and the standby transmission path;

The source node is further configured to record a transmission path response including the primary transmission path and the backup transmission path, and when the primary transmission path fails to transmit the data packet, use the backup transmission path to transmit The datagram.
The data transmission system according to claim 1, characterized in that, the data transmission system further comprises: a first intermediate node, wherein the first intermediate node is located between the source node and the Any node between the destination nodes;

The first intermediate node is configured to determine a fallback path according to the data message;

The first intermediate node is further configured to send a failure notification message to the source node according to the fallback path when detecting a failure of the main transmission path.
The data transmission system according to claim 2, wherein the data transmission system further comprises: one or more second intermediate nodes, wherein the one or more second intermediate nodes are in the main transmission path a node located between said source node and said first intermediate node;

When the first intermediate node determines the fallback path according to the data message, it is specifically used to, according to the ingress port of the one or more second intermediate nodes, the data message carried in the data message , to determine the fallback path.
The data transmission system according to claim 2 or 3, wherein the first intermediate node is further configured to forward the next-hop node to the next-hop node adjacent to the first intermediate node in the main transmission path Before the data message, add the identifier of the ingress port of the data message at the first intermediate node to the data message.
The data transmission system according to claim 4, wherein when the first intermediate node adds the identification of the ingress port of the data message in the data message to the data message, specifically use and replacing the identifier of the egress port of the first intermediate node recorded in the source routing label carried by the data packet with the identifier of the ingress port.
A data transmission method, characterized in that, comprising:

The source node sends a transmission path request to the network manager, and the transmission path request includes the identifier of the destination node;

The source node receives a transmission path response including the primary transmission path and the backup transmission path from the network manager, and records the transmission path response including the primary transmission path and the backup transmission path;

When the data message transmitted by the main transmission path fails, the source node transmits the data message by using the backup transmission path.
The method of claim 6, further comprising:

When the source node receives the failure notification message from the first intermediate node, it determines that the data message transmitted by the main transmission path fails, wherein the first intermediate node is located in the source node on the main transmission path Any node between the node and the destination node.
A data transmission method, characterized in that, comprising:

The network manager receives a transmission path request from the source node, where the transmission path request includes the identifier of the destination node;

The network manager determines a primary transmission path and a backup transmission path between the source node and the destination node;

Before the source node transmits the data packet to the destination node, the network manager sends a transmission path response including the primary transmission path and the standby transmission path to the source node.
A data transmission method, characterized in that, comprising:

The first intermediate node receives the data message from the source node, wherein the first intermediate node is a node located in the main transmission path where the source node and the destination node of the data message transmit the data message;

The first intermediate node determines a fallback path according to the data message;

When detecting that the main transmission path is faulty, the first intermediate node sends a fault notification message to the source node according to the fallback path.
The method according to claim 9, wherein the first intermediate node determines a fallback path according to the data message, comprising:

The first intermediate node determines the fallback path according to the identifiers of the ingress ports of the data message carried in the data message on one or more second intermediate nodes, wherein the one or more first The second intermediate node is a node located between the source node and the first intermediate node in the main transmission path.
The method according to claim 9 or 10, further comprising:

Before forwarding the data message to the next hop node adjacent to the first intermediate node in the main transmission path, the first intermediate node adds the data message in the data message The identifier of the ingress port of the first intermediate node.
A source node, characterized by comprising: a processing unit and a communication unit;

The communication unit is configured to send a transmission path request to a network manager, where the transmission path request includes an identifier of a destination node; and receive a transmission path response from the network manager including a primary transmission path and a backup transmission path;

The processing unit is configured to record a transmission path response including the primary transmission path and the backup transmission path;

The communication unit is further configured to use the standby transmission path to transmit the data packet when the primary transmission path fails to transmit the data packet.
The source node according to claim 12, wherein the processing unit is further configured to determine that the main transmission path transmits a datagram when the communication unit receives a failure notification message from the first intermediate node A file failure occurs, wherein the first intermediate node is any node located between the source node and the destination node in the main transmission path.
A network manager, characterized by comprising: a processing unit and a communication unit;

The communication unit is configured to receive a transmission path request from a source node, where the transmission path request includes an identifier of a destination node;

The processing unit is configured to determine a main transmission path and a backup transmission path between the source node and the destination node;

The communication unit is further configured to send a transmission path response including the primary transmission path and the backup transmission path to the source node before the source node transmits the data message to the destination node.
A first intermediate node, characterized by comprising: a processing unit and a communication unit;

The communication unit is configured to receive a data message from a source node, wherein the first intermediate node is located in the main transmission path where the source node and the destination node of the data message transmit the data message node;

The processing unit is configured to determine a fallback path according to the data message;

The communication unit is further configured to send a failure notification message to the source node according to the fallback path when a failure of the main transmission path is detected.
The first intermediate node according to claim 15, wherein when the processing unit determines the fallback path according to the data message, it is specifically configured to Identifying said fallback path at an ingress port of one or more second intermediate nodes, wherein said one or more second intermediate nodes are between said source node and said first Nodes between intermediate nodes.
The first intermediate node according to claim 15 or 16, wherein the processing unit is further configured to transmit the communication unit to the next next adjacent node in the main transmission path to the first intermediate node. Before forwarding the data message, the hopping node adds an identifier of an ingress port of the data message at the first intermediate node to the data message.
The first intermediate node according to claim 17, wherein when the processing unit adds the identification of the ingress port of the data message in the data message to the data message, it is specifically used to replacing the identifier of the egress port of the first intermediate node recorded in the source routing label carried by the data packet with the identifier of the ingress port.
A data transmission device, characterized in that the device includes a processor and a memory, the memory is used to store computer-executable instructions, and when the data transmission device is running, the processor executes the computer-executable instructions in the memory Instructions are used to execute the operation steps of the method in any one of claims 6-11 by using the hardware resources in the data transmission device.
A computer-readable storage medium, characterized in that a computer program is stored in the computer-readable storage medium, and when the computer program is executed by a computer, the computer executes any one of claims 6-11. Operational steps of the method.