CN119135611A

CN119135611A - Network load balancing control method, device, equipment and storage medium

Info

Publication number: CN119135611A
Application number: CN202310689375.2A
Authority: CN
Inventors: 王瑞雪; 王雯萱; 秦凤伟; 顾戎
Original assignee: China Mobile Communications Group Co Ltd; Research Institute of China Mobile Communication Co Ltd
Current assignee: China Mobile Communications Group Co Ltd; Research Institute of China Mobile Communication Co Ltd
Priority date: 2023-06-12
Filing date: 2023-06-12
Publication date: 2024-12-13

Abstract

The application discloses a control method, a device, equipment and a storage medium for network load balancing. The method comprises the steps of sending a path detection request message to second network equipment by the first network equipment, receiving a plurality of path detection response messages returned by the second network equipment, determining weight values of all link paths based on round trip time RTT values of all path detection response messages, and distributing service messages to corresponding link paths in proportion based on the weight values of all link paths and sending the service messages to the second network equipment. Therefore, the service message can be distributed to the corresponding link paths in proportion based on the weight value of each link path and sent to the second network equipment, so that a single-flow multi-path load sharing mechanism is realized, the service message is transmitted through multi-path distribution, the utilization rate of the link paths in the network is improved, the uniformity of network flow scheduling is optimized, flow congestion is effectively avoided, and the overall performance of the network is improved.

Description

Control method, device, equipment and storage medium for network load balancing

Technical Field

The present application relates to the field of network transmission technologies, and in particular, to a control method, apparatus, device, and storage medium for network load balancing.

Background

In the related art, an RDMA (Remote Direct Memory Access, remote direct data access) lossless network needs to meet requirements of low latency, zero packet loss, high throughput rate and the like, and congestion needs to be avoided as much as possible if zero packet loss is guaranteed. Solving the congestion problem caused by uneven network traffic scheduling is one of the key elements of guaranteeing a lossless network. Therefore, there is a need to optimize network traffic scheduling for lossless networks to meet the network performance requirements of the lossless networks.

Disclosure of Invention

In view of this, the embodiments of the present application provide a method, apparatus, device, and storage medium for controlling network load balancing, so as to effectively improve overall performance of a network.

The technical scheme of the embodiment of the application is realized as follows:

The embodiment of the application provides a control method for network load balancing, which is applied to first network equipment and comprises the following steps:

Transmitting a path detection request message to second network equipment;

receiving a plurality of path detection response messages returned by the second network device, wherein each path detection response message corresponds to a link path connecting the first network device and the second network device, and each path detection response message carries the network address of each node device of the link path;

Determining a weight value of each link path based on an RTT (round-trip time) value of each path detection response message;

And distributing the service message to the corresponding link path in proportion based on the weight value of each link path and sending the service message to the second network equipment.

In the above solution, the determining the weight value of each link path based on the RTT value of each path detection response message includes:

determining a weight value of each link path based on a proportion value of the RTT value of each path detection response message in the sum of RTT values of the plurality of path detection response messages;

wherein the weight value is inversely related to the scale value.

In the above solution, the network address carried by the path detection response message is an IPv6 (Internet Protocol Version, sixth version of internet protocol) address, and the method further includes:

extracting IPv6 addresses of all node devices carried by each path detection response message;

Generating a corresponding SRv6 (Segment Routing IPv, segment routing header) SRH (Segment Routing Header ) extension header of the SRv (Segment Routing IPv, segment routing based on internet protocol version six) based on the IPv6 address of the node device of each of the link path approaches;

the distributing the service message to the corresponding link path according to the proportion and sending the service message to the second network device includes:

And dividing the service message into data packets according to a proportion based on the weight value of each link path, selecting corresponding SRH extension heads from the divided data packets, and packaging the data packets into SRv messages to be sent to the second network equipment, wherein the SRH extension heads of each SRv message are in one-to-one correspondence with the link paths.

In the above scheme, the method further comprises:

Path probe information representing a reachable link path between the first network device and the second network device is constructed based on the plurality of path probe response messages.

In the above scheme, the method further comprises:

sending the path detection request message to the second network equipment based on a set interval duration;

and updating the path detection information and the weight value of each link path based on the received path detection response message returned by the second network equipment.

In the above solution, the path detection information includes a status identifier corresponding to each link path, where the status identifier includes a first identifier indicating that the link path is reachable and a second identifier indicating that the link path is not reachable, and the updating the path detection information includes:

And if the existence of at least one link path with the unreachable path is determined, updating the state identification of the at least one link path to the second identification.

In the above scheme, updating the weight value of each link path includes:

and determining the weight value of the link path of each state identifier as the first identifier according to the RTT value of the path detection response message of the link path of the state identifier as the first identifier.

The embodiment of the application also provides a control device for network load balancing, which is applied to the first network equipment and comprises:

the detection request module is used for sending a path detection request message to the second network equipment;

A receiving module, configured to receive a plurality of path detection response messages returned by the second network device, where each path detection response message corresponds to a link path connecting the first network device and the second network device, and each path detection response message carries a network address of each node device in the link path;

The weight determining module is used for determining the weight value of each link path based on the RTT value of each path detection response message;

and the service processing module is used for distributing the service message to the corresponding link path according to the proportion based on the weight value of each link path and sending the service message to the second network equipment.

The embodiment of the application also provides first network equipment, which comprises a processor and a memory for storing a computer program capable of running on the processor, wherein the processor is used for executing the steps of the method of the embodiment of the application when the computer program runs.

The embodiment of the application also provides a computer storage medium, and the computer storage medium stores a computer program which, when being executed by a processor, realizes the steps of the method of the embodiment of the application.

The technical scheme provided by the embodiment of the application comprises the steps of sending a path detection request message to second network equipment by first network equipment, receiving a plurality of path detection response messages returned by the second network equipment, wherein each path detection response message corresponds to a link path connecting the first network equipment and the second network equipment, each path detection response message carries a network address of each node equipment of a link path, determining a weight value of each link path based on round trip time RTT value of each path detection response message, and distributing service messages to corresponding link paths in proportion based on the weight value of each link path and sending the service messages to the second network equipment. Therefore, the service message can be distributed to the corresponding link paths in proportion based on the weight value of each link path and sent to the second network equipment, so that a single-flow multi-path load sharing mechanism is realized, the service message is transmitted through multi-path distribution, the utilization rate of the link paths in the network is improved, the uniformity of network flow scheduling is optimized, flow congestion is effectively avoided, and the overall performance of the network is improved.

Drawings

Fig. 1 is a schematic diagram of a switching principle between a primary link and a backup link in a load sharing mechanism based on a per-flow in the related art;

Fig. 2 is a schematic diagram of an application of a load sharing mechanism based on a per-flow in the related art;

fig. 3 is a flow chart of a control method of network load balancing according to an embodiment of the present application;

fig. 4 is a schematic diagram of converting path detection information into SRv SRH extension header in an application example of the present application;

Fig. 5 is a schematic structural diagram of SRv a message in an application example of the present application;

Fig. 6 is a schematic diagram of SRv to message transmission in an application example of the present application;

fig. 7 is a schematic structural diagram of a control device for network load balancing according to an embodiment of the present application;

Fig. 8 is a schematic structural diagram of a network device according to an embodiment of the present application.

Detailed Description

The present application will be described in further detail with reference to the accompanying drawings and examples.

Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein in the description of the application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application.

Related RDMA lossless networks generally employ RC (Reliable Connection ) mode, so that a main research direction for solving the problem of uneven network traffic scheduling is achieved by a load sharing mechanism based on a stream-by-stream basis. For example, the flow scheduling between the server network card and the switch adopts ECMP (Equal-Cost Multi-Path routing) which is a load sharing mechanism based on a per-flow basis. When ECMP load sharing is used, the network card takes five-tuple (source IP, destination IP, source port number, destination port number and protocol number) of the data packet as a HASH factor, generates a HASH-KEY value through a HASH algorithm, and then selects a member link from load sharing links according to the HASH-KEY value to forward the data packet. When the five-tuple is the same, the network card always selects the same next-hop address as the last time to send the message. When the five-tuple is different, the network card can select a relatively idle path for forwarding.

As shown in fig. 1, when a link1 link fails, the route on the original link1 link is quickly converged by IGP (Interior Gateway Protocol ), and a new link (such as link2 link) is selected to retransmit the traffic. After the link1 link is restored to be normal, the network card can reselect the link1 link to send data because the priority of the link1 link is highest. Thus, a switch back from the backup link to the primary link is achieved.

However, the above-described flow-by-flow-based load sharing mechanism is essentially a single-flow single-path traffic scheduling approach. This scheduling method, although the principle is simple to implement, lacks a mechanism for considering whether or not there is congestion on the path itself. In actual service, although the flow hash is uniform, the load of each equivalent link is not uniform, so that the overall network performance cannot be improved.

For the data packet with the same five-tuple characteristic field, the HASH-KEY value is the same, so that the same member link is selected for forwarding, and the condition that a plurality of flows are hashed to the same path easily occurs, so that the link is overloaded, a certain physical link is overloaded, and even congestion occurs, so that the message is discarded. As shown in fig. 2, although the flow2 applies the ECMP function to disperse the traffic with the 30G bandwidth to different links for forwarding, when a link is selected, it is not considered that an elephant flow exists on one member link, so that congestion occurs, and at the same time, the load of other links is lighter, and the bandwidth utilization rate is low. In a data center in a high-performance computing and distributed storage application scenario, there is both a latency-sensitive mouse stream and an elephant stream with high bandwidth throughput. When an elephant stream and a mouse stream are statically hashed to the same link, the elephant stream can jam and squeeze the mouse stream, resulting in serious degradation of application performance.

Based on this, in various embodiments of the present application, a control method for network load balancing based on a load sharing mechanism of single-flow multipath is provided, and the load balancing performance of the control method is obviously better than that of the load sharing mechanism based on the per-flow, so that the defect of traffic congestion caused by static hash of the elephant flow and the mouse flow onto the same link can be effectively avoided, and further, the performance requirements of low time delay, zero packet loss, high throughput rate and the like are really satisfied.

As shown in fig. 3, an embodiment of the present application provides a control method for network load balancing, which is applied to a first network device, and the control method includes:

Step 301, sending a path measurement probe request message to the second network device.

Here, the first network device may be a transmitting network card, and the second network device may be a receiving network card.

The first network device sends a path detection request message with a destination address being the network address of the second network device, the path detection request message carries a detection mark, the node devices along the way determine that the path detection request message is based on the detection mark, then add own network address into the path detection request message, find all outgoing interfaces reaching the destination address of the path detection request message in a routing table, and send the path detection request message from the interfaces until the path detection request message is sent to the second network device.

Step 302, receiving a plurality of path detection response messages returned by the second network device, where each path detection response message corresponds to a link path connecting the first network device and the second network device, and each path detection response message carries a network address of each node device of the link path.

It will be appreciated that after receiving the path probe request messages transmitted through the aforementioned interfaces, the second network device may return each path probe request message (i.e. a path probe response message) to the first network device along the original path, so that the first network device will receive path probe response messages of all reachable paths.

It can be understood that each path detection response message corresponds to a link path, and each path detection response message carries the network address of each node device of the link path.

Step 303, determining a weight value of each link path based on the RTT value of each path detection response message.

Here, the first network device may determine the weight value of the corresponding link path based on the RTT value of each path detection response packet, so that the traffic weight of each link path may be intelligently allocated, and the control effect of load balancing is improved.

For example, the first network device may calculate the RTT value based on the sending time of the path probe request message and the arrival time of the path probe response message recorded by the timestamp in the header of the path probe response message, where RRT value=arrival time-sending time.

Illustratively, the determining the weight value of each link path based on the RTT value of each path detection response message includes:

wherein the weight value is inversely related to the scale value.

Here, the negative correlation of the weight value with the scale value means that the weight value decreases as the scale value increases. In an application example, assuming that the first network device receives m path detection response messages, an RTT value of an i-th path detection response message is ti, and a sum of RTT values of the m path detection response messages is totaled, where i e (1, m) is a natural number greater than 1, a weight value Ω of a link path corresponding to the i-th path detection response message may be expressed as Ω=1- (ti/totaled), so that an RTT value based on the path detection response message may be obtained, and a weight value of each link path may be determined.

Step 304, based on the weight value of each link path, distributing the service message to the corresponding link path proportionally and sending to the second network device.

Here, because the first network device determines the weight value of each link path based on the foregoing step 303, a packet-by-packet load sharing manner may be adopted to distribute the data packet of the service packet to the corresponding link path according to the weight value of each link path and send the data packet to the second network device, so as to implement a load sharing mechanism of single-flow multipath, and transmit the service packet through multipath splitting, which is beneficial to improving the utilization rate of the link paths in the network, optimizing the uniformity of network traffic scheduling, effectively avoiding traffic congestion, and further improving the overall performance of the network.

It should be noted that, in the control method of the embodiment of the present application, a load sharing mechanism based on single-flow and multi-path is adopted, so that compared with a load sharing mechanism based on stream by stream, the defect of traffic congestion caused by static hash of an elephant flow and a mouse flow on the same link can be effectively overcome, and thus performance requirements such as low delay, zero packet loss, high throughput rate and the like can be practically satisfied.

Illustratively, the network address carried by the path probe response message is an IPv6 address, and the method further includes:

generating a corresponding SRv SRH extension header based on the IPv6 address of the node equipment of each link path;

It can be understood that after the first network device receives the service packet, the service packet may be sliced to a fixed length or a variable length, for example, an MTU (Maximum Transmission Unit, the maximum transmission unit) is adopted to slice the service packet, and the number of data packets with a corresponding proportion is allocated according to the weight value of each link path, that is, the data packets sliced by the MTU are allocated to each link path in proportion, and before the data packets are forwarded, packet header encapsulation is performed based on the SRH extension header corresponding to the link path, and the packet header encapsulation is performed to obtain SRv packet, and the SRv packet is sent to the second network device, so that multipath split transmission of the service packet is realized, and traffic congestion is effectively avoided.

It should be noted that SRv is a Segment route transmission technology based on IPv6 data plane forwarding, where a source node designates a path for a service packet, converts the path into an ordered Segment list, encapsulates the Segment list into an SRH extension header, and an intermediate node of the path only needs to forward according to the path designated in the SRH extension header, so that the network is simplified and has good scalability.

In an application example, assume that the relation between the IPv6 address set pathm = { SID [ n ], SID [ n-1],..and SID [0] }, pathm and SRv SRH extension header of the mth link path node device is shown in fig. 4.

Illustratively, a SRv packet of a SRv SRH extension header corresponding to a link path sent by a data packet is added to the data packet as shown in fig. 5.

Referring to fig. 4 and 6, in SRv a message, the IPv6 destination address field is a variable field whose value is determined by both the SEGMENT LEFT locator field and the Segment list function field. For example, when SEGMENT LEFT points to an active field SEGMENT LIST [2], the IPv6 address of SEGMENT LIST [2] needs to be copied to the IPv6 destination address field, the IPv6 routing table is searched for and forwarded, and SEGMENT LEFT is updated and decremented, then the pointer is biased to the new active field, when SEGMENT LEFT field is decremented to 0, it indicates that SRv message is delivered to the destination node, the destination node pops up the SRH extension header, and then the message is processed in the next step.

The first network device may distribute the service packet to each link path according to a proportion, and may further distribute a sequence number to the data packet of each link path, so that after receiving SRv messages, the second network device unpacks the packet, may perform out-of-order rearrangement based on the sequence number of the data packet, and further restore the service packet to obtain the service packet, upload the service packet to an application, a network card, or a CPU (central processing unit) for processing, so as to meet a processing requirement of a subsequent service packet.

Illustratively, the control method according to the embodiment of the present application further includes:

In an application example, the first network device may construct a path probe information maintenance table in which path probe information for each link path between the first network device and the second network device, e.g., an IPv6 address set of the link path routing node device, is stored.

It can be appreciated that the first network device can quickly construct the SRv SRH extension header corresponding to the link path based on the path detection information.

It can be understood that the first network device may periodically send the path detection request packet to the second network device, so that the path detection information and the weight value of each link path may be dynamically updated, which is beneficial to continuously optimizing and improving the network performance.

Illustratively, the path probing information includes a status identifier corresponding to each link path, the status identifier including a first identifier indicating that the link path is reachable and a second identifier indicating that the link path is not reachable, the updating the path probing information including:

In an application example, the path detection information maintenance table further includes a status identifier of the path detection information, where if the link path is not reachable, the status identifier may be set to idle, and if the link path is reachable, the status identifier may be set to active.

After sending out the path detection request message, the first network device determines that the historical link path does not receive the path detection response message within a set duration, and then sets the state identifier of the path detection information of the link path as idle.

Illustratively, updating the weight value of each of the link paths includes:

It can be understood that the first network device only updates the respective weight values for the link paths with the active state identifier, that is, when the first network device performs multipath forwarding on the service packet, deletes the link paths with the idle state identifier (that is, cancels the load sharing qualification of the link paths), updates the weight values of the link paths reachable by each path based on the latest RTT value, divides the service packet according to the weight values of each link path, selects the corresponding SRH extension header from the divided data packet, encapsulates the data packet into SRv packets, and sends the SRv packets to the second network device, so that the dynamic update of the weight values of the link paths and the dynamic allocation of the per-packet load sharing are realized, which is beneficial to practically guaranteeing the overall performance of the network.

In order to implement the method of the embodiment of the present application, the embodiment of the present application further provides a control device for network load balancing, where the control device for network load balancing corresponds to the control method for network load balancing, and each step in the embodiment of the control method for network load balancing is also completely applicable to the embodiment of the control device for network load balancing.

The control device for network load balancing is applied to a first network device, and as shown in fig. 7, the control device for network load balancing comprises a probe request module 701, a receiving module 702, a weight determining module 703 and a service processing module 704. The system comprises a detection request module 701, a receiving module 702, a weight determining module 703 and a service processing module 704, wherein the detection request module 701 is used for sending a path detection request message to a second network device, the receiving module 702 is used for receiving a plurality of path detection response messages returned by the second network device, each path detection response message corresponds to a link path connecting the first network device and the second network device, each path detection response message carries a network address of each node device of the link path, the weight determining module 703 is used for determining a weight value of each link path based on an RTT value of each path detection response message, and the service processing module 704 is used for distributing a service message to the corresponding link path in proportion and sending the service message to the second network device based on the weight value of each link path.

In some embodiments, the weight determining module 703 determines a weight value of each link path based on an RTT value of each path probe response message, including:

wherein the weight value is inversely related to the scale value.

In some embodiments, the network address carried by the path probe response packet is an IPv6 address, and the service processing module 704 is further configured to:

In some embodiments, the weight determination module 703 is further configured to:

In some embodiments, the probe request module 701 is further configured to send the path probe request packet to the second network device based on a set interval duration, and the weight determining module 703 is further configured to update the path probe information and the weight value of each link path based on the received path probe response packet returned by the second network device.

In some embodiments, the path probing information includes a status identifier corresponding to each link path, the status identifier including a first identifier indicating that the link path is reachable and a second identifier indicating that the link path is not reachable, the updating the path probing information including:

In some embodiments, updating the weight value of each of the link paths includes:

In actual application, the probe request module 701, the receiving module 702, the weight determining module 703 and the service processing module 704 may be implemented by a processor in the first network device. Of course, the processor needs to run a computer program in memory to implement its functions.

It should be noted that, when the control device for network load balancing provided in the foregoing embodiment performs control of network load balancing, only the division of the foregoing program modules is used as an example, in practical application, the foregoing process allocation may be performed by different program modules according to needs, that is, the internal structure of the device is divided into different program modules to complete all or part of the foregoing processes. In addition, the control device for network load balancing provided in the above embodiment and the control method embodiment for network load balancing belong to the same concept, and detailed implementation processes of the control device for network load balancing are shown in the method embodiment, and are not repeated here.

Based on the hardware implementation of the program modules, and in order to implement the method of the embodiment of the present application, the embodiment of the present application further provides a network device. The network device (i.e., the aforementioned first network device) may be a transmitting network card. Fig. 8 shows only an exemplary structure of the network device, not all of which may be implemented as needed.

As shown in fig. 8, a network device 800 provided by an embodiment of the present application includes at least one processor 801, memory 802, a user interface 803, and at least one network interface 804. The various components in network device 800 are coupled together by a bus system 805. It is appreciated that the bus system 805 is used to enable connected communications between these components. The bus system 805 includes a power bus, a control bus, and a status signal bus in addition to the data bus. But for clarity of illustration, the various buses are labeled as bus system 805 in fig. 8.

The user interface 803 may include, among other things, a display, keyboard, mouse, trackball, click wheel, keys, buttons, touch pad, or touch screen, etc.

The memory 802 in embodiments of the present application is used to store various types of data to support the operation of the network device. Examples of such data include any computer program for operating on a network device.

The control method for network load balancing disclosed by the embodiment of the application can be applied to the processor 801 or realized by the processor 801. The processor 801 may be an integrated circuit chip with signal processing capabilities. In implementation, the steps of the control method for network load balancing may be performed by integrated logic circuits of hardware in the processor 801 or instructions in the form of software. The Processor 801 may be a general purpose Processor, a digital signal Processor (DSP, digital Signal Processor), or other programmable logic device, discrete gate or transistor logic device, discrete hardware components, or the like. The processor 801 may implement or perform the methods, steps, and logic blocks disclosed in embodiments of the present application. The general purpose processor may be a microprocessor or any conventional processor or the like. The steps of the method disclosed in the embodiment of the application can be directly embodied in the hardware of the decoding processor or can be implemented by combining hardware and software modules in the decoding processor. The software modules may be located in a storage medium, such as a memory 802, where the processor 801 reads information from the memory 802, and in combination with its hardware, performs the steps of the control method for network load balancing provided by the embodiments of the present application.

In an exemplary embodiment, the network device may be implemented by one or more Application Specific Integrated Circuits (ASICs), DSPs, programmable logic devices (PLDs, programmable Logic Device), complex programmable logic devices (CPLDs, complex Programmable Logic Device), field programmable gate arrays (FPGAs, field Programmable GATE ARRAY), general purpose processors, controllers, microcontrollers (MCUs, micro Controller Unit), microprocessors (microprocessors), or other electronic elements for performing the aforementioned methods.

It is to be appreciated that memory 802 can be either volatile memory or nonvolatile memory, and can include both volatile and nonvolatile memory. The non-volatile Memory may be, among other things, a Read Only Memory (ROM), a programmable Read Only Memory (PROM, programmable Read-Only Memory), erasable programmable Read-Only Memory (EPROM, erasable Programmable Read-Only Memory), electrically erasable programmable Read-Only Memory (EEPROM, ELECTRICALLY ERASABLE PROGRAMMABLE READ-Only Memory), Magnetic random access Memory (FRAM, ferromagnetic random access Memory), flash Memory (Flash Memory), magnetic surface Memory, optical disk, or compact disk-Only Memory (CD-ROM, compact Disc Read-Only Memory), which may be disk Memory or tape Memory. the volatile memory may be random access memory (RAM, random Access Memory) which acts as external cache memory. By way of example and not limitation, many forms of RAM are available, such as static random access memory (SRAM, static Random Access Memory), synchronous static random access memory (SSRAM, synchronous Static Random Access Memory), dynamic random access memory (DRAM, dynamic Random Access Memory), synchronous dynamic random access memory (SDRAM, synchronous Dynamic Random Access Memory), and, Double data rate synchronous dynamic random access memory (DDRSDRAM, double Data Rate Synchronous Dynamic Random Access Memory), enhanced synchronous dynamic random access memory (ESDRAM, enhanced Synchronous Dynamic Random Access Memory), synchronous link dynamic random access memory (SLDRAM, syncLink Dynamic Random Access Memory), Direct memory bus random access memory (DRRAM, direct Rambus Random Access Memory). The memory described by embodiments of the present application is intended to comprise, without being limited to, these and any other suitable types of memory.

In an exemplary embodiment, the present application further provides a computer storage medium, which may be a computer readable storage medium, for example, including a memory 802 storing a computer program, where the computer program may be executed by the processor 801 of the network device to perform the steps described in the method of the embodiment of the present application. The computer readable storage medium may be ROM, PROM, EPROM, EEPROM, flash Memory, magnetic surface Memory, optical disk, or CD-ROM.

It should be noted that "first," "second," etc. are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order.

In addition, the embodiments of the present application may be arbitrarily combined without any collision.

The foregoing is merely illustrative of the present application, and the present application is not limited thereto, and any person skilled in the art will readily recognize that variations or substitutions are within the scope of the present application. Therefore, the protection scope of the application is subject to the protection scope of the claims.

Claims

1. A method for controlling network load balancing, characterized in that it is applied to a first network device, comprising:

Sending a path detection request message to the second network device;

receiving a plurality of path detection response messages returned by the second network device, wherein each of the path detection response messages corresponds to a link path connecting the first network device and the second network device, and each of the path detection response messages carries a network address of each node device along the link path;

Determine a weight value of each link path based on the round trip time (RTT) value of each path detection response message;

Based on the weight value of each link path, the service message is proportionally distributed to the corresponding link path and sent to the second network device.

2. The method according to claim 1, characterized in that the determining the weight value of each link path based on the RTT value of each path detection response message comprises:

Determine a weight value of each of the link paths based on a ratio of the RTT value of each of the path detection response messages to the sum of the RTT values of the multiple path detection response messages;

The weight value is negatively correlated with the ratio value.

3. The method according to claim 1, wherein the network address carried by the path detection response message is an IPv6 address, and the method further comprises:

Extracting the IPv6 addresses of all node devices carried in each of the path detection response messages;

Generate a corresponding SRv6 segment routing header (SRH) extension header based on the IPv6 address of the node device along each link path;

The allocating the service message to the corresponding link path in proportion and sending the service message to the second network device includes:

The service message is divided into data packets in proportion based on the weight value of each link path, and the divided data packets are selected with corresponding SRH extension headers, encapsulated into SRv6 messages and sent to the second network device, wherein the SRH extension headers of each SRv6 message correspond one-to-one to the link path.

4. The method according to claim 1, characterized in that the method further comprises:

Path detection information indicating a link path reachable between the first network device and the second network device is constructed based on the multiple path detection response messages.

5. The method according to claim 4, characterized in that the method further comprises:

Sending the path detection request message to the second network device based on a set interval duration;

Based on the received path detection response message returned by the second network device, the path detection information and the weight value of each link path are updated.

6. The method according to claim 5, characterized in that the path detection information includes a state identifier corresponding to each link path, the state identifier includes a first identifier indicating that the link path is reachable and a second identifier indicating that the link path is unreachable, and updating the path detection information includes:

If it is determined that there is at least one link path that is unreachable, the state identifier of the at least one link path is updated to the second identifier.

7. The method according to claim 6, characterized in that updating the weight value of each link path comprises:

According to the RTT value of the path detection response message of the link path whose state identifier is the first identifier, the weight value of each link path whose state identifier is the first identifier is determined.

8. A network load balancing control device, characterized in that it is applied to a first network device, comprising:

A detection request module, used for sending a path detection request message to the second network device;

A receiving module, configured to receive multiple path detection response messages returned by the second network device, wherein each of the path detection response messages corresponds to a link path connecting the first network device and the second network device, and each of the path detection response messages carries the network address of each node device along the link path;

A weight determination module, used to determine the weight value of each link path based on the RTT value of each path detection response message;

The service processing module is used to distribute the service message to the corresponding link path in proportion based on the weight value of each link path and send it to the second network device.

9. A first network device, comprising: a processor and a memory for storing a computer program that can be run on the processor, wherein:

The processor is used to execute the steps of the method according to any one of claims 1 to 7 when running a computer program.

10. A computer storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.