Traffic scheduling method and device and related equipment

Abstract

The invention relates to the technical field of intelligent calculation centers, in particular to a traffic scheduling method and device and related equipment. The method is applied to a controller, and comprises the following steps: obtaining whole network topology information and host ARP information, the whole network topology information comprising each link relationship and access interface information included by each link; the initial overhead value of each link included in the whole network topology information is defined, and the size of the initial overhead value of one link is inversely proportional to the maximum bandwidth value of the link; source and target end information of a to-be-trained task is determined, a target path used for transmitting the traffic of the to-be-trained task is determined based on the source and target end information of the to-be-trained task, host ARP information and overhead values of all links included in the whole network topology information, and target links included in the target path are used for transmitting the traffic of the to-be-trained task. The sum of the overhead values of the target links is smaller than or equal to a preset value.

Description

Technical Field

[0001] The present application relates to the technical field of intelligent computing centers, and in particular to a traffic scheduling method, device and related equipment. Background Art

[0002] In the intelligent computing data center network, the upstream and downstream bandwidth is 1:1, and there will be many equal-cost routing (ECMP, Equal-Cost Multi-Path routing) links in the network. In the case of flow-by-flow forwarding, the switch will perform hash routing forwarding based on the five-tuple of the traffic and other calculation factors to achieve load balancing in the network. The traffic from multiple entrances is hashed ECMP to the same exit. There is no traffic on other exits, resulting in unbalanced load sharing and network congestion. In other words, in the intelligent computing data center network, the switch equipment hashes or loads unevenly, resulting in network traffic congestion that cannot be solved. Summary of the invention

[0003] The present application provides a traffic scheduling method, device and related equipment.

[0004] In a first aspect, the present application provides a traffic scheduling method, which is applied to a controller, and the method includes:

[0005] Obtaining network topology information of the entire network and host ARP information, wherein the network topology information of the entire network includes the relationship between each link and the ingress and egress interface information of each link;

[0006] Defining an initial cost value of each link included in the network topology information of the entire network, wherein the size of the initial cost value of a link is inversely proportional to the maximum bandwidth value of the link;

[0007] Determine the source and destination information of the task to be trained, and based on the source and destination information of the task to be trained, the host ARP information and the cost value of each link included in the network topology information of the whole network, determine the target path for transmitting the traffic of the task to be trained, wherein the target path includes the target links, and the cost value of the target links is less than or equal to a preset value.

[0008] Optionally, the controller maintains a current cost value of each link; after determining a target path for transmitting the task traffic to be trained, the method further includes:

[0009] The current cost value of each target link included in the target path is updated.

[0010] Optionally, the step of updating the current cost value of each target link included in the target path includes:

[0011] The current cost value of each target link included in the target path is accumulated by 1.

[0012] Optionally, the step of determining source destination information of the task to be trained, and determining a target path for transmitting the traffic of the task to be trained based on the source destination information of the task to be trained, the host ARP information, and the cost value of each link included in the network topology information of the whole network includes:

[0013] Determine the first access device accessed by the source end and the second access device accessed by the destination end according to the source destination end information and the host ARP information;

[0014] Determine at least one path between the first access device and the second access device according to the network topology information of the entire network;

[0015] Calculate the path cost value of the at least one path respectively according to the cost value of each link included in the network topology information of the whole network;

[0016] According to the path cost value of the at least one path, a path whose path cost value is less than or equal to a preset value is determined as a target path for transmitting the task traffic to be trained.

[0017] Optionally, the method further comprises:

[0018] When it is detected that the training of the task to be trained is completed and the target path is released, the current cost value of each target link included in the target path is cumulatively reduced by 1.

[0019] In a second aspect, the present application provides a flow scheduling device, applied to a controller, the device comprising:

[0020] An acquisition unit, used to acquire network topology information of the entire network and host ARP information, wherein the network topology information of the entire network includes the relationship between each link and the ingress and egress interface information of each link;

[0021] A definition unit, used to define an initial cost value of each link included in the network topology information of the entire network, wherein the size of the initial cost value of a link is inversely proportional to the maximum bandwidth value of the link;

[0022] A determination unit is used to determine the source destination information of the task to be trained, and based on the source destination information of the task to be trained, the host ARP information and the cost value of each link included in the network topology information of the whole network, determine the target path for transmitting the traffic of the task to be trained, wherein the target path includes the target links, and the cost value of the target links is less than or equal to a preset value.

[0023] Optionally, the controller maintains a current cost value of each link; and the device further includes:

[0024] An updating unit is used to update the current cost value of each target link included in the target path.

[0025] Optionally, when updating the current cost value of each target link included in the target path, the updating unit is further used to:

[0026] The current cost value of each target link included in the target path is accumulated by 1.

[0027] Optionally, when determining the source destination information of the task to be trained, and determining the target path for transmitting the traffic of the task to be trained based on the source destination information of the task to be trained, the host ARP information and the cost value of each link included in the network topology information of the whole network, the determining unit is specifically used to:

[0028] Determine the first access device accessed by the source end and the second access device accessed by the destination end according to the source destination end information and the host ARP information;

[0029] Determine at least one path between the first access device and the second access device according to the network topology information of the entire network;

[0030] Calculate the path cost value of the at least one path respectively according to the cost value of each link included in the network topology information of the whole network;

[0031] According to the path cost value of the at least one path, a path whose path cost value is less than or equal to a preset value is determined as a target path for transmitting the task traffic to be trained.

[0032] Optionally, the device further comprises a detection unit:

[0033] When the detection unit detects that the training of the task to be trained is completed and the target path is released, the updating unit is further configured to decrement the current cost value of each target link included in the target path by 1.

[0034] In a third aspect, an embodiment of the present application provides a traffic scheduling device, the traffic scheduling device comprising:

[0035] A memory for storing program instructions;

[0036] The processor is used to call the program instructions stored in the memory, and execute the steps of the method as described in any one of the first aspects above according to the obtained program instructions.

[0037] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to enable the computer to execute the steps of the method described in any one of the first aspects above.

[0038] In summary, the traffic scheduling method provided in the embodiment of the present application is applied to a controller, and the method includes: obtaining network topology information and host ARP information of the entire network, wherein the network topology information of the entire network includes the relationship between each link and the input and output interface information included in each link; defining the initial cost value of each link included in the network topology information of the entire network, wherein the size of the initial cost value of a link is inversely proportional to the maximum bandwidth value of the link; determining the source and destination information of the task to be trained, and based on the source and destination information of the task to be trained, the host ARP information and the cost value of each link included in the network topology information of the entire network, determining the target path for transmitting the traffic of the task to be trained, wherein the target path includes the target link, and the sum of the cost values ​​of the target link is less than or equal to a preset value.

[0039] Using the traffic scheduling method provided in the embodiment of the present application, the controller maintains the cost value of each link in the entire network topology. When selecting a route, it calculates the total cost value of each path and selects the path with the smallest total cost value as the forwarding path for the training traffic currently to be scheduled, thereby avoiding network congestion caused by uneven equipment load. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments of the present application or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For ordinary technicians in this field, other drawings can also be obtained based on these drawings of the embodiments of the present application.

[0041] Figure 1 is a flow scheduling schematic diagram;

[0042] Figure 2 A detailed flow chart of a traffic scheduling method provided in an embodiment of the present application;

[0043] Figure 3 A schematic diagram of the structure of a flow scheduling device provided in an embodiment of the present application;

[0044] Figure 4 A schematic diagram of the hardware architecture of a traffic scheduling device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0045] The terms used in the embodiments of the present application are only for the purpose of describing specific embodiments, rather than limiting the present application. The singular forms of "a", "said" and "the" used in the present application and claims are also intended to include plural forms, unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used herein refers to any or all possible combinations of one or more associated listed items.

[0046] It should be understood that, although the terms first, second, third, etc. may be used to describe various information in the embodiments of the present application, these information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of the present application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, in addition, the word "if" used may be interpreted as "at..." or "when..." or "in response to determination".

[0047] In the intelligent computing data center network, the upstream and downstream bandwidth is 1:1, and there will be many equal-cost routing (ECMP, Equal-Cost Multi-Path routing) links in the network. In the case of flow-by-flow forwarding, the switch device will perform hash routing forwarding based on the five-tuple of the traffic and other calculation factors to achieve load balancing in the network. The traffic of multiple entrances is hashed ECMP to the same exit. If there is no traffic on other exits, it will lead to unbalanced load sharing and network congestion. As follows Figure 1 As shown in the figure, the red links indicate that two groups of training tasks (node1 and node2, node3 and node4) are scheduled to the same spine and the same outbound interface. During training, when the traffic is loaded to the same Leaf as shown in the figure above, the Spine may be loaded to the same path (outbound interface) when going to Leaf2, resulting in uneven traffic load and congestion in spine routing.

[0048] That is, currently in the intelligent computing data center networking, the switch equipment hash or load is uneven, resulting in network traffic congestion that cannot be solved. The embodiment of the present application provides a solution to solve the network congestion caused by uneven load by planning the forwarding path of the entire network through a controller.

[0049] For example, see Figure 2 , which is a detailed flow chart of a traffic scheduling method provided in an embodiment of the present application, the method is applied to a controller, and the method includes the following steps:

[0050] Step 200: Acquire network topology information of the entire network and host ARP information, wherein the network topology information of the entire network includes the relationship between each link and the input and output interface information included in each link.

[0051] In an embodiment of the present application, the controller manages network devices in the entire network. Through the Link Layer Discovery Protocol (LLDP), the controller can normally obtain the network topology of the entire network and the host ARP (Address Resolution Protocol) information, while recording the link relationship and input and output interface information of each link included in the network topology of the entire network. The controller can perform a baseline processing on this topology, that is, save the link information (link relationship, input and output interface information of each link) and the host ARP information in the memory. In this way, the controller has a global perspective of the topology relationship and host ARP information.

[0052] In actual applications, the controller can obtain information about the specific access device interface to which the server (node, such as a network card of a server) is connected through the host ARP information. The training task traffic refers to the traffic between two nodes (such as node1->node2). In this way, the access device to which node1 is connected (a specific interface of the access device) and the specific interface of the access device to which node2 is connected can be obtained based on the host ARP information.

[0053] Step 210: define the initial cost value of each link included in the network topology information of the entire network.

[0054] The initial cost value of a link is inversely proportional to the maximum bandwidth value of the link.

[0055] In the embodiment of the present application, the Spine-Leaf two-layer network architecture is used as an example for description, and the network includes Spine1, Spine2, Leaf-1, Leaf-2 and Leaf-3, where:

[0056] Uplink interface 1 of Leaf-1 is connected to downlink interface 1 of Spine1, uplink interface 2 is connected to downlink interface 2 of Spine1, uplink interface 3 of Leaf-1 is connected to downlink interface 1 of Spine2, and uplink interface 4 is connected to downlink interface 1 of Spine2.

[0057] Uplink interface 1 of Leaf-2 is connected to downlink interface 3 of Spine1, uplink interface 2 is connected to downlink interface 4 of Spine1, uplink interface 3 of Leaf-2 is connected to downlink interface 3 of Spine2, and uplink interface 4 is connected to downlink interface 4 of Spine2;

[0058] Uplink interface 1 of Leaf-3 is connected to downstream interface 5 of Spine1, and upstream interface 2 is connected to downstream interface 6 of Spine1. Uplink interface 3 of Leaf-3 is connected to downstream interface 5 of Spine2, and upstream interface 4 is connected to downstream interface 6 of Spine2.

[0059] The connections between the upstream interfaces of the leaf devices and the downstream interfaces of the spine devices are defined as links.

[0060] When the network is initialized, the cost value of the link needs to be defined according to the link quality of each link (e.g., the maximum available bandwidth of the link). Preferably, if the maximum available bandwidth of each link is the same, the cost value of each link can be defined as equal (e.g., defined as 1). That is, the cost value (current cost value) of each link maintained in the memory of the controller is 1.

[0061] Step 220: Determine the source destination information of the task to be trained, and determine the target path for transmitting the traffic of the task to be trained based on the source destination information of the task to be trained, the host ARP information and the cost value of each link included in the network topology information of the entire network.

[0062] The target path includes a target link, and the sum of the cost values ​​of the target link is less than or equal to a preset value.

[0063] In an embodiment of the present application, when determining the source destination information of the task to be trained, and determining the target path for transmitting the traffic of the task to be trained based on the source destination information of the task to be trained, the host ARP information and the cost value of each link included in the network topology information of the whole network, a preferred implementation method is:

[0064] Determine the first access device accessed by the source end and the second access device accessed by the destination end according to the source destination end information and the host ARP information;

[0065] Determine at least one path between the first access device and the second access device according to the network topology information of the entire network;

[0066] Calculate the path cost value of the at least one path respectively according to the cost value of each link included in the network topology information of the whole network;

[0067] According to the path cost value of the at least one path, a path whose path cost value is less than or equal to a preset value is determined as a target path for transmitting the task traffic to be trained.

[0068] In an embodiment of the present application, before starting a training task, the platform can call the interface of the controller to import five-tuple information such as the source and destination of the training task (source and destination information). In this way, the controller can determine the first access device accessed by the source and the second device accessed by the destination through the host ARP information; then, the controller determines at least one forwarding path from the first device to the second device (each link included in each forwarding path) through the network topology information; finally, based on the current cost value of each link maintained locally, the controller determines the forwarding path with the smallest total cost value of each link as the target path for transmitting the training task traffic.

[0069] For example, assuming that the forwarding path from the first device to the second device includes path 1 and path 2, path 1 includes link 1 and link 3, path 2 includes link 2 and link 4, and the total overhead value of link 1 and link 3 included in path 1 is 1+1=2; and the total overhead value of link 2 and link 4 included in path 2 is 1+3=4, then path 1 is determined as the target path for forwarding training task traffic.

[0070] In actual applications, the controller sends a high priority forwarding flow table to each network device included in the target path, so that each network device forwards the traffic of the task to be trained based on the forwarding flow table. After the training of the task to be trained is completed, each network device is instructed to delete the corresponding forwarding flow table.

[0071] In the embodiment of the present application, since the controller maintains the current cost value of each link, after determining the target path for transmitting the task traffic to be trained, the traffic scheduling method may further include the following steps:

[0072] The current cost value of each target link included in the target path is updated.

[0073] Specifically, when updating the current cost value of each target link included in the target path, a preferred implementation method is:

[0074] The current cost value of each target link included in the target path is accumulated by 1.

[0075] In the following, we take path 1 (including link 1 and link 3) as the target path for transmitting training task traffic as an example. The current cost value of link 1 is 1, and the current cost value of link 3 is also 1. At this time, the controller accumulates 1 to the cost value of link 1 and updates it to 2; similarly, the controller accumulates 1 to the current cost value of link 3 and updates it to 2.

[0076] In practical applications, when updating the current cost value of a link, a fixed value can be accumulated, or a custom setting can be made according to the traffic size of the latest training task carried. For example, if the traffic of the latest training task carried is large, then when updating the current cost value of the corresponding link, the accumulated value is also large. For example, if the training task traffic is less than a certain value, 1 is accumulated, and if the training task traffic is greater than a certain value, 2 is accumulated. In the embodiments of the present application, the specific implementation method is not limited.

[0077] Furthermore, in an embodiment of the present application, when it is detected that the training of the task to be trained is completed and the target path is released, the current overhead value of each target link included in the target path is reduced by 1.

[0078] That is, after a training task is completed and the path (links) corresponding to the training task is released, the current overhead value of each link must be subtracted from the accumulated value when carrying the training task. For example, if the accumulated value is 1, the subtracted value is also 1.

[0079] Based on the same inventive concept as the above-mentioned embodiment of the invention, for example, refer to Figure 3 FIG. 1 is a schematic diagram of a flow scheduling device provided in an embodiment of the present application. The device is applied to a controller and includes:

[0080] The acquisition unit 30 is used to acquire network topology information of the entire network and host ARP information, wherein the network topology information of the entire network includes the relationship between each link and the ingress and egress interface information of each link;

[0081] A definition unit 31, configured to define an initial cost value of each link included in the network topology information of the entire network, wherein the size of the initial cost value of a link is inversely proportional to the maximum bandwidth value of the link;

[0082] The determination unit 32 is used to determine the source destination information of the task to be trained, and based on the source destination information of the task to be trained, the host ARP information and the cost value of each link included in the network topology information of the whole network, determine the target path for transmitting the traffic of the task to be trained, wherein the target path includes the target links, and the cost value of the target links is less than or equal to a preset value.

[0083] Optionally, the controller maintains a current cost value of each link; and the device further includes:

[0084] An updating unit is used to update the current cost value of each target link included in the target path.

[0085] Optionally, when updating the current cost value of each target link included in the target path, the updating unit is further used to:

[0086] The current cost value of each target link included in the target path is accumulated by 1.

[0087] Optionally, when determining the source destination information of the task to be trained, and determining the target path for transmitting the traffic of the task to be trained based on the source destination information of the task to be trained, the host ARP information and the cost value of each link included in the network topology information of the whole network, the determining unit 32 is specifically used to:

[0088] Determine the first access device accessed by the source end and the second access device accessed by the destination end according to the source destination end information and the host ARP information;

[0089] Determine at least one path between the first access device and the second access device according to the network topology information of the entire network;

[0090] Calculate the path cost value of the at least one path respectively according to the cost value of each link included in the network topology information of the whole network;

[0091] According to the path cost value of the at least one path, a path whose path cost value is less than or equal to a preset value is determined as a target path for transmitting the task traffic to be trained.

[0092] Optionally, the device further comprises a detection unit:

[0093] When the detection unit detects that the training of the task to be trained is completed and the target path is released, the updating unit is further configured to decrement the current cost value of each target link included in the target path by 1.

[0094] The above units may be one or more integrated circuits configured to implement the above methods, such as one or more application specific integrated circuits (ASIC), or one or more digital signal processors (DSP), or one or more field programmable gate arrays (FPGA). For another example, when a certain unit is implemented in the form of a processing element scheduling program code, the processing element may be a general-purpose processor, such as a central processing unit (CPU) or other processor that can call program code. For another example, these units may be integrated together and implemented in the form of a system-on-a-chip (SOC).

[0095] Furthermore, the flow scheduling device provided in the embodiment of the present application, from the hardware level, the hardware architecture diagram of the flow scheduling device can be seen in Figure 4 As shown, the traffic scheduling device may include: a memory 40 and a processor 41,

[0096] The memory 40 is used to store program instructions; the processor 41 calls the program instructions stored in the memory 40 and executes the above method embodiment according to the obtained program instructions. The specific implementation method and technical effect are similar and will not be repeated here.

[0097] Optionally, the present application also provides a controller, comprising at least one processing element (or chip) for executing the above method embodiment.

[0098] Optionally, the present application also provides a program product, such as a computer-readable storage medium, which stores computer-executable instructions, and the computer-executable instructions are used to enable the computer to execute the above method embodiments.

[0099] Here, the machine-readable storage medium may be any electronic, magnetic, optical or other physical storage device that may contain or store information, such as executable instructions, data, etc. For example, the machine-readable storage medium may be: RAM (RadomAccess Memory), volatile memory, non-volatile memory, flash memory, storage drive (such as hard disk drive), solid state drive, any type of storage disk (such as CD, DVD, etc.), or similar storage medium, or a combination thereof.

[0100] The systems, devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer, which may be in the form of a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email transceiver, a game console, a tablet computer, a wearable device or a combination of any of these devices.

[0101] For the convenience of description, the above device is described in various units according to their functions. Of course, when implementing the present application, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0102] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the embodiments of the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0103] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0104] Moreover, these computer program instructions can also be stored in a computer-readable memory that can guide a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a product including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0105] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable device to implement the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0106] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A traffic scheduling method, characterized in that: Applied to a controller, the method comprises: Obtaining network topology information of the entire network and host ARP information, wherein the network topology information of the entire network includes the relationship between each link and the ingress and egress interface information of each link; Defining an initial cost value of each link included in the network topology information of the entire network, wherein the size of the initial cost value of a link is inversely proportional to the maximum bandwidth value of the link; Determine the source and destination information of the task to be trained, and based on the source and destination information of the task to be trained, the host ARP information and the cost value of each link included in the network topology information of the whole network, determine the target path for transmitting the traffic of the task to be trained, wherein the target path includes the target links, and the cost value of the target links is less than or equal to a preset value.

2. The method according to claim 1, characterized in that The controller maintains a current cost value of each link; After determining the target path for transmitting the task traffic to be trained, the method further includes: The current cost value of each target link included in the target path is updated.

3. The method according to claim 2, characterized in that The step of updating the current cost value of each target link included in the target path includes: The current cost value of each target link included in the target path is accumulated by 1.

4. The method according to any one of claims 1 to 3, characterized in that: The steps of determining source destination information of the task to be trained, and determining a target path for transmitting the traffic of the task to be trained based on the source destination information of the task to be trained, the host ARP information, and the cost value of each link included in the network topology information of the whole network include: Determine the first access device accessed by the source end and the second access device accessed by the destination end according to the source destination end information and the host ARP information; Determine at least one path between the first access device and the second access device according to the network topology information of the entire network; Calculate the path cost value of the at least one path respectively according to the cost value of each link included in the network topology information of the whole network; According to the path cost value of the at least one path, a path whose path cost value is less than or equal to a preset value is determined as a target path for transmitting the task traffic to be trained.

5. The method according to claim 3, characterized in that The method further comprises: When it is detected that the training of the task to be trained is completed and the target path is released, the current cost value of each target link included in the target path is cumulatively reduced by 1.

6. A flow scheduling device, characterized in that: Applied to a controller, the device comprises: An acquisition unit, used to acquire network topology information of the entire network and host ARP information, wherein the network topology information of the entire network includes the relationship between each link and the ingress and egress interface information of each link; A definition unit, used to define an initial cost value of each link included in the network topology information of the entire network, wherein the size of the initial cost value of a link is inversely proportional to the maximum bandwidth value of the link; A determination unit is used to determine the source destination information of the task to be trained, and based on the source destination information of the task to be trained, the host ARP information and the cost value of each link included in the network topology information of the whole network, determine the target path for transmitting the traffic of the task to be trained, wherein the target path includes the target links, and the cost value of the target links is less than or equal to a preset value.

7. The device according to claim 6, characterized in that The controller maintains the current cost value of each link; the device also includes: An updating unit is used to update the current cost value of each target link included in the target path.

8. The device according to claim 6 or 7, characterized in that When determining the source destination information of the task to be trained, and determining the target path for transmitting the traffic of the task to be trained based on the source destination information of the task to be trained, the host ARP information, and the cost value of each link included in the network topology information of the whole network, the determining unit is specifically used to: Determine the first access device accessed by the source end and the second access device accessed by the destination end according to the source destination end information and the host ARP information; Determine at least one path between the first access device and the second access device according to the network topology information of the entire network; Calculate the path cost value of the at least one path respectively according to the cost value of each link included in the network topology information of the whole network; According to the path cost value of the at least one path, a path whose path cost value is less than or equal to a preset value is determined as a target path for transmitting the task traffic to be trained.

9. A flow scheduling device, characterized in that: The flow scheduling device comprises: A memory for storing program instructions; A processor is used to call the program instructions stored in the memory, and execute the steps of the method according to any one of claims 1 to 5 according to the obtained program instructions.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to enable the computer to execute the steps of the method according to any one of claims 1 to 5.

Images