CN114095364B - Network congestion control method and device - Google Patents
Network congestion control method and device Download PDFInfo
- Publication number
- CN114095364B CN114095364B CN202111432632.1A CN202111432632A CN114095364B CN 114095364 B CN114095364 B CN 114095364B CN 202111432632 A CN202111432632 A CN 202111432632A CN 114095364 B CN114095364 B CN 114095364B
- Authority
- CN
- China
- Prior art keywords
- scene
- ecn
- configuration
- parameters
- ecn configuration
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 34
- 238000003062 neural network model Methods 0.000 claims description 23
- 230000006870 function Effects 0.000 claims description 21
- 230000002787 reinforcement Effects 0.000 claims description 15
- 238000012360 testing method Methods 0.000 claims description 13
- 238000012549 training Methods 0.000 claims description 13
- 238000010276 construction Methods 0.000 claims description 3
- 230000008569 process Effects 0.000 description 8
- 230000003068 static effect Effects 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/10—Flow control; Congestion control
- H04L47/26—Flow control; Congestion control using explicit feedback to the source, e.g. choke packets
- H04L47/263—Rate modification at the source after receiving feedback
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/08—Configuration management of networks or network elements
- H04L41/0803—Configuration setting
- H04L41/0823—Configuration setting characterised by the purposes of a change of settings, e.g. optimising configuration for enhancing reliability
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/14—Network analysis or design
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The application provides a network congestion control method and device. The method comprises the following steps: acquiring scene parameters of an outlet to be controlled in network equipment in a current scene, wherein the scene parameters are parameters related to flow forwarded through the outlet to be controlled; inputting scene parameters in a current scene and the ECN configuration into a trained flow model aiming at each preset ECN configuration to obtain a network performance value corresponding to the ECN configuration in the current scene, wherein the network performance value is used for identifying network performance which can be achieved by executing congestion control based on the ECN configuration; selecting ECN configuration corresponding to the maximum network performance value as the optimal ECN configuration in the current scene; and carrying out network congestion control on the to-be-controlled output port based on the optimal ECN configuration. It can be seen that the dynamic ECN configuration can be realized, so that the dynamic ECN configuration is adapted to the flow scenes with dynamic changes, and the network performance under each flow scene is improved.
Description
Technical Field
The present disclosure relates to the field of communications technologies, and in particular, to a method and an apparatus for controlling network congestion.
Background
With the increase of high-concurrency and low-delay traffic, network congestion easily occurs at the output ports of network devices (such as switches and routers). Network congestion refers to a phenomenon that when a network device receives a much larger flow through an ingress port than a flow sent through an egress port, a large amount of messages are retained (buffered) in an egress queue (corresponding to the egress port), which affects network performance such as transmission delay and throughput of the messages.
Currently, static display congestion notification (English: explicit Congestion Notification, abbreviated: ECN) configurations are commonly employed to control network congestion. Specifically, the network device judges whether the queue is congested according to the ECN threshold configured statically, if congestion occurs, adds ECN identification to the message in the queue, sends the message to the destination device, and after receiving the message with ECN identification, the destination device sends a congestion notification message (English: congestion Notification Packet, abbreviated: CNP) to the source device, and notifies the source device to reduce the traffic sending rate so as to achieve the purpose of controlling network congestion.
However, static ECN configuration cannot adapt to dynamically changing traffic scenarios, resulting in poor network performance in each traffic scenario.
Disclosure of Invention
In view of this, the present application proposes a network congestion control method and device, which are used to effectively improve network performance in each traffic scenario while implementing network congestion control.
In order to achieve the purposes of the application, the application provides the following technical scheme:
in a first aspect, the present application provides a network congestion control method, where the method includes:
acquiring scene parameters of an outlet to be controlled in network equipment in a current scene, wherein the scene parameters are parameters related to flow forwarded through the outlet to be controlled;
inputting scene parameters in the current scene and the ECN configuration into a trained flow model aiming at each preset ECN configuration to obtain a network performance value corresponding to the ECN configuration in the current scene, wherein the network performance value is used for identifying network performance which can be achieved by executing congestion control based on the ECN configuration;
selecting ECN configuration corresponding to the maximum network performance value as the optimal ECN configuration in the current scene;
and carrying out network congestion control on the to-be-controlled output port based on the optimal ECN configuration.
Optionally, the inputting the scene parameters in the current scene and the ECN configuration into the trained traffic model includes:
matching a preset configuration file with the scene parameters in the current scene, wherein the configuration file is used for recording the corresponding relation between the scene parameters and ECN configuration;
and if the scene parameters in the current scene are not hit in any scene parameters in the configuration file, inputting the scene parameters in the current scene and the ECN configuration selected currently into a trained flow model.
Optionally, the method further comprises:
and if the scene parameters in the current scene hit the target scene parameters in the configuration file, the ECN configuration corresponding to the target scene parameters is used as the optimal ECN configuration in the current scene.
Optionally, before the matching of the scene parameters in the current scene with the preset configuration file, the method further includes:
and adding the corresponding relation between the scene parameters in the common scene and the optimal ECN configuration in the common scene obtained through testing into the configuration file for each common scene.
Optionally, before the inputting the scene parameters in the current scene and the ECN configuration into the trained traffic model, the method further includes:
constructing a reinforcement learning model, wherein a cost function of the reinforcement learning model is used for representing network performances expected to be achieved under different scenes and different ECN configurations;
constructing a neural network model for expressing the cost function, wherein the input of the neural network model is scene parameters and ECN configuration, and the output of the neural network model is a network performance value;
and training the neural network model by utilizing scene parameters in various common scenes and various preset ECN configurations to obtain a flow model which is enough to approximate the network performance expected by the cost function.
In a second aspect, the present application provides a network congestion control apparatus, the apparatus comprising:
the device comprises an acquisition unit, a control unit and a control unit, wherein the acquisition unit is used for acquiring scene parameters of an outlet to be controlled in network equipment in a current scene, wherein the scene parameters are parameters related to flow forwarded through the outlet to be controlled;
the input unit is used for inputting the scene parameters in the current scene and the ECN configuration into the trained flow model aiming at each preset ECN configuration to obtain a network performance value corresponding to the ECN configuration in the current scene, wherein the network performance value is used for identifying network performance which can be achieved by executing congestion control based on the ECN configuration;
the selection unit is used for selecting ECN configuration corresponding to the maximum network performance value as the optimal ECN configuration in the current scene;
and the control unit is used for controlling the network congestion of the to-be-controlled output port based on the optimal ECN configuration.
Optionally, the input unit inputs the scene parameters in the current scene and the ECN configuration into a trained traffic model, including:
matching a preset configuration file with the scene parameters in the current scene, wherein the configuration file is used for recording the corresponding relation between the scene parameters and ECN configuration;
and if the scene parameters in the current scene are not hit in any scene parameters in the configuration file, inputting the scene parameters in the current scene and the ECN configuration selected currently into a trained flow model.
Optionally, the selecting unit is further configured to, if the scene parameter in the current scene hits the target scene parameter in the configuration file, use the ECN configuration corresponding to the target scene parameter as the optimal ECN configuration in the current scene.
Optionally, the apparatus further includes:
and the adding unit is used for adding the corresponding relation between the scene parameters in the common scene and the optimal ECN configuration in the common scene obtained through testing into the configuration file for each common scene.
Optionally, the apparatus further includes:
a building unit, configured to build a reinforcement learning model, where a cost function of the reinforcement learning model is used to represent network performance expected to be achieved under different scenarios and different ECN configurations;
the construction unit is further used for constructing a neural network model for expressing the cost function, wherein the input of the neural network model is scene parameters and ECN configuration, and the output of the neural network model is a network performance value;
and the training unit is used for training the neural network model by utilizing scene parameters in various common scenes and various preset ECN configurations to obtain a flow model which is close to the network performance expected by the cost function.
As can be seen from the above description, in the embodiment of the present application, by using the generalization capability of the trained model (flow model), the network performance that can be achieved under the corresponding ECN configuration is accurately estimated for different scenarios and different ECN configurations, and the ECN configuration that can optimize the network performance under the current scenario is selected to perform congestion control. Therefore, the method can adapt to dynamically-changed traffic scenes, and better network performance can be obtained under each traffic scene.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are needed in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flowchart of a network congestion control method according to an embodiment of the present application;
FIG. 2 is a flow chart illustrating an implementation of step 102 in accordance with an embodiment of the present application;
FIG. 3 is a flow model training process shown in an embodiment of the present application;
FIG. 4 is a reinforcement learning model framework shown in an embodiment of the present application;
fig. 5 is a schematic structural diagram of a network congestion control apparatus according to an embodiment of the present application.
Detailed Description
Reference will now be made in detail to exemplary embodiments, examples of which are illustrated in the accompanying drawings. When the following description refers to the accompanying drawings, the same numbers in different drawings refer to the same or similar elements, unless otherwise indicated. The implementations described in the following exemplary examples are not representative of all implementations consistent with the present application.
The terminology used in the embodiments of the application is for the purpose of describing particular embodiments only and is not intended to be limiting of the embodiments of the application. As used in the embodiments of the present application, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should also be understood that the term "and/or" as used herein refers to and encompasses any or all possible combinations of one or more of the associated listed items.
It should be understood that although the terms first, second, third, etc. may be used in embodiments of the present application to describe various information, these information should not be limited to these terms. These terms are only used to distinguish one type of information from another. For example, the negotiation information may also be referred to as second information, and similarly, the second information may also be referred to as negotiation information, without departing from the scope of embodiments of the present application. The word "if" as used herein may be interpreted as "at … …" or "at … …" or "responsive to a determination", depending on the context.
For the purposes, technical solutions and advantages of the present application, the following detailed description of the present application is described with reference to the accompanying drawings and specific embodiments:
referring to fig. 1, a flowchart of a network congestion control method is shown in an embodiment of the present application, where the flowchart may be applied to a network device to perform congestion control, for example, a switch, a router, and the like, and may also be applied to a control device independent from the network device, for example, a server, and the like.
The network device includes at least one egress port. The administrator may specify the egress ports for which congestion control needs to be performed, or perform congestion control on all egress ports by default. Here, the output ports that need to perform congestion control are all referred to as output ports to be controlled.
The flow shown in fig. 1 is performed for each output port to be controlled. As shown in fig. 1, the process may include the steps of:
step 101, obtaining a scene parameter of an output port to be controlled in a current scene, wherein the scene parameter is a parameter related to the flow forwarded through the output port to be controlled.
Here, the scene refers to a traffic scene at a certain time. The scene parameter refers to a flow related parameter under a certain scene, and in this application, refers to a parameter of the flow forwarded through the output port to be controlled. In the following description, if not specified, the traffic refers to the traffic forwarded through the output port to be controlled.
Specifically, the scene parameters may include: average bandwidth of traffic source ports (ingress ports), number of flows, proportion of different types of messages, etc. Here, it should be noted that forwarding traffic through the same egress port may come from one or more ingress ports.
Step 102, inputting the scene parameters in the current scene and the ECN configuration into the trained flow model for each preset ECN configuration, to obtain a network performance value corresponding to the ECN configuration in the current scene, where the network performance value is used to identify network performance that can be achieved by executing congestion control based on the ECN configuration.
In this embodiment of the present application, a plurality of selectable ECN configurations may be preset, and in subsequent processing, an ECN configuration that is most suitable for a current scenario needs to be selected from the plurality of ECN configurations.
Here, it should be noted that one ECN configuration may include at least the following three parameters: an ECN threshold upper limit, an ECN threshold lower limit, and an ECN marking probability (maximum marking probability).
To better understand these three parameters, a brief description of the process of congestion control based on ECN configuration is provided below.
When the dequeue depth is lower than the ECN threshold lower limit, the network equipment does not add ECN marks to the messages in the dequeue; when the dequeue depth is higher than the ECN threshold upper limit, adding ECN marks for all messages in the dequeue; messages in the dequeue are randomly marked according to a linear probability (between 0 and a maximum marking probability) determined based on the dequeue depth when the dequeue depth is between the lower ECN threshold limit and the upper ECN threshold limit.
In this step, for each preset ECN configuration, the scene parameters of the current scene and the ECN configuration are input into a trained traffic model, and the traffic model outputs the network performance that can be achieved by congestion control based on the ECN configuration in the current scene. In particular, the achievable network performance may be identified by outputting a network performance value that identifies the network performance. Wherein, the larger the network performance value is, the better the corresponding network performance is represented.
It can be seen that in the embodiment of the present application, the traffic model is a pre-trained model for measuring the influence of ECN configuration on network performance. Where network performance may include a combination of one or more of throughput, latency, etc.
Through the step, the network performance value corresponding to each preset ECN configuration in the current scene can be obtained.
And step 103, selecting ECN configuration corresponding to the maximum network performance value as the optimal ECN configuration in the current scene.
That is, an ECN configuration is selected that optimizes network performance in the current scenario.
And 104, performing network congestion control on the to-be-controlled output port based on the optimal ECN configuration.
Specifically, if the flow shown in fig. 1 is executed by the network device, in this step, the network device may directly execute network congestion control on the local to-be-controlled output port according to the optimal ECN configuration determined in step 103; if the control device executes the flow shown in fig. 1, in this step, the control device needs to issue the optimal ECN configuration determined in step 103 to the network device, and then the network device executes network congestion control on the local egress port to be controlled based on the issued ECN configuration. Of course, if the optimal ECN configuration determined by the control device is the same as an existing ECN configuration corresponding to the to-be-controlled output port in the network device, the control device may not issue the optimal ECN configuration.
Thus, the flow shown in fig. 1 is completed.
As can be seen from the flow shown in fig. 1, in the embodiment of the present application, by using the generalization capability of the trained model (flow model), the network performance that can be achieved under the corresponding ECN configuration is accurately estimated for different scenarios and different ECN configurations, and the ECN configuration that can optimize the network performance under the current scenario is selected to perform congestion control. Therefore, the method can adapt to dynamically-changed traffic scenes, and better network performance can be obtained under each traffic scene.
The process of inputting the scene parameters and ECN configuration in the current scene into the trained traffic model is described below in step 102.
Referring to fig. 2, a flow of implementation of step 102 is shown in an embodiment of the present application. As shown in fig. 2, the process may include the steps of:
step 201, matching a preset configuration file with a scene parameter in a current scene, where the configuration file is used to record a corresponding relationship between the scene parameter and the ECN configuration.
Here, it should be noted that the configuration file may be generated according to test results for several common scenarios.
As an example, a test environment for testing different common scenarios may be built in a laboratory. For each common scene, testing network performance which can be achieved when congestion control is carried out by adopting different preset ECN configurations in the common scene, selecting ECN configuration which can enable the network performance to be optimal in the common scene as ECN configuration of the current common scene, and adding the corresponding relation between scene parameters of the common scene and the optimal ECN configuration in the common scene obtained by testing into a configuration file. Namely, the corresponding relation between the scene parameters of the common scene and the optimal ECN configuration in the common scene is stored through the configuration file.
In this step, the scene parameters of the current scene obtained in step 101 are matched with the scene parameters in the configuration file.
Step 202, if the scene parameters in the current scene miss any scene parameters in the configuration file, inputting the scene parameters in the current scene and the ECN configuration selected currently into the trained flow model.
If any one of the scene parameters in the configuration file is missed by the scene parameters in the current scene, which indicates that the current scene does not belong to the tested common scene, inputting the scene parameters in the current scene and the selected preset ECN configuration into a trained flow model, and estimating the network performance which can be achieved by congestion control based on the preset ECN configuration in the current scene through the flow model, wherein the description of step 102 is omitted.
Otherwise, if the scene parameters in the current scene hit the target scene parameters in the configuration file, the current scene is indicated to belong to the tested common scene, the ECN configuration corresponding to the target scene parameters is directly obtained from the configuration file, the ECN configuration is used as the optimal ECN configuration in the current scene, and network congestion control is performed on the ports to be controlled based on the optimal ECN configuration.
Thus, the flow shown in fig. 2 is completed.
As can be seen from the flow shown in fig. 2, in the embodiment of the present application, for a common scene, an optimal ECN configuration in the common scene can be quickly obtained by matching a configuration file, so as to ensure congestion control efficiency in the common scene; for unusual scenes (the unusual scenes have a large number), the generalization capability of the traffic model can be used to estimate the network performance under different ECN configurations, so as to select the ECN configuration suitable for the current unusual scene.
The flow model training process is described below.
Referring to fig. 3, a flow model training process is shown in an embodiment of the present application. As shown in fig. 3, the process may include the steps of:
in step 301, a reinforcement learning model is constructed, and a cost function of the reinforcement learning model is used to represent network performance expected to be achieved under different ECN configurations in different scenarios.
Referring to fig. 4, a reinforcement learning model framework is shown in an embodiment of the present application.
In this embodiment of the present application, the "environment" included in the framework may be a scene parameter under a specific scene, for example, an average bandwidth of each ingress port, the number of flows, and a ratio of different types of messages; the "state" may be the counted number of messages added with ECN marks, throughput of dequeues, dequeue depth, current ECN configuration, etc.; the action is the configuration of the issued ECN; "rewards" are added values of network performance.
Wherein, the Value (Value) function of the reinforcement learning model can be used to represent the network performance expected to be achieved under different ECN configurations in different scenes.
In step 302, a neural network model for expressing the cost function is constructed, wherein the inputs of the neural network model are scene parameters and ECN configuration, and the outputs are network performance values.
Step 303, training a neural network model by using scene parameters in each common scene and each preset ECN configuration to obtain a flow model which is enough to approximate the network performance expected by the cost function.
In this step, model training is performed using test data of each common scenario (including scenario parameters of the common scenario, different preset ECN configurations) to obtain a flow model that sufficiently approximates the network performance expected by the cost function. The network performance expected by the cost function may be measured network performance obtained when testing for different common scenarios and different ECN configurations.
Thus, the flow shown in fig. 3 is completed.
As can be seen from the flow shown in fig. 3, in the embodiment of the present application, the flow model is trained by using the test data in the common scenario, so that the network performance in different ECN configurations in any scenario can be accurately estimated, thereby laying a good foundation for selecting the optimal ECN configuration based on the estimated result in the following steps (step 103 and step 104).
The method provided by the embodiment of the present application is described above, and the device provided by the embodiment of the present application is described below:
referring to fig. 5, a network congestion control apparatus according to an embodiment of the present application includes an obtaining unit 501, an input unit 502, a selecting unit 503, and a control unit 504, where:
an obtaining unit 501, configured to obtain a scene parameter of an output port to be controlled in a network device in a current scene, where the scene parameter is a parameter related to a flow forwarded through the output port to be controlled;
an input unit 502, configured to input, for each preset ECN configuration, a scene parameter in the current scene and the ECN configuration into a trained traffic model, to obtain a network performance value corresponding to the ECN configuration in the current scene, where the network performance value is used to identify network performance that can be achieved by performing congestion control based on the ECN configuration;
a selecting unit 503, configured to select an ECN configuration corresponding to the maximum network performance value as an optimal ECN configuration in the current scenario;
and a control unit 504, configured to perform network congestion control on the to-be-controlled output port based on the optimal ECN configuration.
As an embodiment, the input unit 502 inputs the scene parameters in the current scene and the ECN configuration into a trained traffic model, including:
matching a preset configuration file with the scene parameters in the current scene, wherein the configuration file is used for recording the corresponding relation between the scene parameters and ECN configuration;
and if the scene parameters in the current scene are not hit in any scene parameters in the configuration file, inputting the scene parameters in the current scene and the ECN configuration selected currently into a trained flow model.
As an embodiment, the selecting unit 503 is further configured to, if the scene parameter in the current scene hits the target scene parameter in the configuration file, use the ECN configuration corresponding to the target scene parameter as the optimal ECN configuration in the current scene.
As an embodiment, the apparatus further comprises:
and the adding unit is used for adding the corresponding relation between the scene parameters in the common scene and the optimal ECN configuration in the common scene obtained through testing into the configuration file for each common scene.
As an embodiment, the apparatus further comprises:
a building unit, configured to build a reinforcement learning model, where a cost function of the reinforcement learning model is used to represent network performance expected to be achieved under different scenarios and different ECN configurations;
the construction unit is further used for constructing a neural network model for expressing the cost function, wherein the input of the neural network model is scene parameters and ECN configuration, and the output of the neural network model is a network performance value;
and the training unit is used for training the neural network model by utilizing scene parameters in various common scenes and various preset ECN configurations to obtain a flow model which is close to the network performance expected by the cost function.
The description of the apparatus shown in fig. 5 is thus completed.
As can be seen from the above description, in the embodiment of the present application, by using the generalization capability of the trained model (flow model), the network performance that can be achieved under the corresponding ECN configuration is accurately estimated for different scenarios and different ECN configurations, and the ECN configuration that can optimize the network performance under the current scenario is selected to perform congestion control. Therefore, the method can adapt to dynamically-changed traffic scenes, and better network performance can be obtained under each traffic scene.
The foregoing description of the preferred embodiments is merely exemplary in nature and is not intended to limit the invention to the precise form disclosed, and thus, any modification, equivalents, and alternatives falling within the spirit and scope of the embodiments are intended to be included within the scope of the invention.
Claims (10)
1. A method for controlling network congestion, the method comprising:
acquiring scene parameters of an outlet to be controlled in network equipment in a current scene, wherein the scene parameters are parameters related to flow forwarded through the outlet to be controlled;
inputting scene parameters in the current scene and the ECN configuration into a trained flow model aiming at each preset congestion notification ECN configuration to obtain a network performance value corresponding to the ECN configuration in the current scene, wherein the network performance value is used for identifying network performance which can be achieved by executing congestion control based on the ECN configuration;
selecting ECN configuration corresponding to the maximum network performance value as the optimal ECN configuration in the current scene;
and carrying out network congestion control on the to-be-controlled output port based on the optimal ECN configuration.
2. The method of claim 1, wherein said inputting the scene parameters in the current scene and the ECN configuration into the trained traffic model comprises:
matching a preset configuration file with the scene parameters in the current scene, wherein the configuration file is used for recording the corresponding relation between the scene parameters and ECN configuration;
and if the scene parameters in the current scene are not hit in any scene parameters in the configuration file, inputting the scene parameters in the current scene and the ECN configuration selected currently into a trained flow model.
3. The method of claim 2, wherein the method further comprises:
and if the scene parameters in the current scene hit the target scene parameters in the configuration file, the ECN configuration corresponding to the target scene parameters is used as the optimal ECN configuration in the current scene.
4. A method according to claim 2 or 3, wherein before said matching a preset profile with scene parameters in said current scene, the method further comprises:
and adding the corresponding relation between the scene parameters in the common scene and the optimal ECN configuration in the common scene obtained through testing into the configuration file for each common scene.
5. The method of claim 4, wherein before inputting the scene parameters in the current scene and the ECN configuration into the trained traffic model, the method further comprises:
constructing a reinforcement learning model, wherein a cost function of the reinforcement learning model is used for representing network performances expected to be achieved under different scenes and different ECN configurations;
constructing a neural network model for expressing the cost function, wherein the input of the neural network model is scene parameters and ECN configuration, and the output of the neural network model is a network performance value;
and training the neural network model by utilizing scene parameters in various common scenes and various preset ECN configurations to obtain a flow model which is enough to approximate the network performance expected by the cost function.
6. A network congestion control apparatus, the apparatus comprising:
the device comprises an acquisition unit, a control unit and a control unit, wherein the acquisition unit is used for acquiring scene parameters of an outlet to be controlled in network equipment in a current scene, wherein the scene parameters are parameters related to flow forwarded through the outlet to be controlled;
the input unit is used for inputting the scene parameters in the current scene and the ECN configuration into the trained flow model for each preset congestion notification ECN configuration to obtain a network performance value corresponding to the ECN configuration in the current scene, wherein the network performance value is used for identifying network performance which can be achieved by executing congestion control based on the ECN configuration;
the selection unit is used for selecting ECN configuration corresponding to the maximum network performance value as the optimal ECN configuration in the current scene;
and the control unit is used for controlling the network congestion of the to-be-controlled output port based on the optimal ECN configuration.
7. The apparatus of claim 6, wherein the input unit inputs the scene parameters in the current scene and the ECN configuration into a trained traffic model, comprising:
matching a preset configuration file with the scene parameters in the current scene, wherein the configuration file is used for recording the corresponding relation between the scene parameters and ECN configuration;
and if the scene parameters in the current scene are not hit in any scene parameters in the configuration file, inputting the scene parameters in the current scene and the ECN configuration selected currently into a trained flow model.
8. The apparatus of claim 7, wherein:
the selecting unit is further configured to, if the scene parameter in the current scene hits the target scene parameter in the configuration file, use the ECN configuration corresponding to the target scene parameter as the optimal ECN configuration in the current scene.
9. The apparatus of claim 7 or 8, wherein the apparatus further comprises:
and the adding unit is used for adding the corresponding relation between the scene parameters in the common scene and the optimal ECN configuration in the common scene obtained through testing into the configuration file for each common scene.
10. The apparatus of claim 9, wherein the apparatus further comprises:
a building unit, configured to build a reinforcement learning model, where a cost function of the reinforcement learning model is used to represent network performance expected to be achieved under different scenarios and different ECN configurations;
the construction unit is further used for constructing a neural network model for expressing the cost function, wherein the input of the neural network model is scene parameters and ECN configuration, and the output of the neural network model is a network performance value;
and the training unit is used for training the neural network model by utilizing scene parameters in various common scenes and various preset ECN configurations to obtain a flow model which is close to the network performance expected by the cost function.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111432632.1A CN114095364B (en) | 2021-11-29 | 2021-11-29 | Network congestion control method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111432632.1A CN114095364B (en) | 2021-11-29 | 2021-11-29 | Network congestion control method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114095364A CN114095364A (en) | 2022-02-25 |
| CN114095364B true CN114095364B (en) | 2024-02-27 |
Family
ID=80305556
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111432632.1A Active CN114095364B (en) | 2021-11-29 | 2021-11-29 | Network congestion control method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114095364B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112511448A (en) * | 2019-09-16 | 2021-03-16 | 华为技术有限公司 | Method for processing network congestion, method for updating model and related device |
| WO2021052374A1 (en) * | 2019-09-16 | 2021-03-25 | 华为技术有限公司 | Network congestion control method, node, system and storage medium |
| WO2021052162A1 (en) * | 2019-09-17 | 2021-03-25 | 华为技术有限公司 | Network parameter configuration method and apparatus, computer device, and storage medium |
| EP3840299A1 (en) * | 2018-09-25 | 2021-06-23 | Huawei Technologies Co., Ltd. | Method and system for collecting statistics about the number of active flows, and network device |
| CN113328953A (en) * | 2020-02-28 | 2021-08-31 | 华为技术有限公司 | Method, device and storage medium for network congestion adjustment |
| CN113572694A (en) * | 2020-04-29 | 2021-10-29 | 华为技术有限公司 | Congestion control method, device and system and computer storage medium |
-
2021
- 2021-11-29 CN CN202111432632.1A patent/CN114095364B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP3840299A1 (en) * | 2018-09-25 | 2021-06-23 | Huawei Technologies Co., Ltd. | Method and system for collecting statistics about the number of active flows, and network device |
| CN112511448A (en) * | 2019-09-16 | 2021-03-16 | 华为技术有限公司 | Method for processing network congestion, method for updating model and related device |
| WO2021052374A1 (en) * | 2019-09-16 | 2021-03-25 | 华为技术有限公司 | Network congestion control method, node, system and storage medium |
| WO2021052162A1 (en) * | 2019-09-17 | 2021-03-25 | 华为技术有限公司 | Network parameter configuration method and apparatus, computer device, and storage medium |
| CN113328953A (en) * | 2020-02-28 | 2021-08-31 | 华为技术有限公司 | Method, device and storage medium for network congestion adjustment |
| CN113572694A (en) * | 2020-04-29 | 2021-10-29 | 华为技术有限公司 | Congestion control method, device and system and computer storage medium |
| JP2021175194A (en) * | 2020-04-29 | 2021-11-01 | 華為技術有限公司Huawei Technologies Co., Ltd. | Congestion control method, apparatus and system, and computer storage medium |
Non-Patent Citations (2)
| Title |
|---|
| intelligent active queue management using explicit congestion notification;cesar A gomez 等;《2019 IEEE global communications conference》;全文 * |
| 改进ABC算法优化LSSVM的网络流量预测模型;许爱军;;计算机应用与软件(第01期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114095364A (en) | 2022-02-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Jay et al. | A deep reinforcement learning perspective on internet congestion control | |
| CN113348645B (en) | System and method for classifying data streams | |
| CN105827472B (en) | Network data flow type detection method and device | |
| Jay et al. | Internet congestion control via deep reinforcement learning | |
| US8751619B2 (en) | Autonomous performance probing | |
| EP3761562A1 (en) | Indicator determination method and related device thereto | |
| CN112532409A (en) | Network parameter configuration method and device, computer equipment and storage medium | |
| US9432295B2 (en) | Systems and methods for high throughput traffic pattern generation | |
| WO2023056808A1 (en) | Encrypted malicious traffic detection method and apparatus, storage medium and electronic apparatus | |
| US7864816B1 (en) | Integrated circuit for network delay and jitter testing | |
| CN116545936A (en) | Congestion control method, system, device, communication equipment and storage medium | |
| US20030067878A1 (en) | Method and system for traffic management in packet networks using random early marking | |
| CN114095364B (en) | Network congestion control method and device | |
| US20050117513A1 (en) | Flow generation method for internet traffic measurement | |
| Li et al. | A fast traffic classification method based on SDN network | |
| Han et al. | Design and performance evaluation of enhanced congestion control algorithm for wireless TCP by using a deep learning | |
| CN105099799A (en) | Botnet detection method and controller | |
| CN115914115B (en) | Network congestion control method, device and communication system | |
| CN114513408B (en) | ECN threshold configuration method and device | |
| CN108023766B (en) | Automatic QoE perception management device based on SDN | |
| CN113938292B (en) | Vulnerability attack flow detection method and detection system based on concept drift | |
| RU2728948C1 (en) | Method for early detection of occurrence moment of poisson's teletraffic overload | |
| CN114070795A (en) | Congestion threshold determination method, device, equipment and storage medium | |
| CN114513409B (en) | ECN threshold configuration method and device | |
| Raida et al. | Lightweight detection of tariff limits in cellular mobile networks |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |