WO2023036042A1 - 网络优化策略的确定方法、装置及系统 - Google Patents

网络优化策略的确定方法、装置及系统 Download PDF

Info

Publication number
WO2023036042A1
WO2023036042A1 PCT/CN2022/116441 CN2022116441W WO2023036042A1 WO 2023036042 A1 WO2023036042 A1 WO 2023036042A1 CN 2022116441 W CN2022116441 W CN 2022116441W WO 2023036042 A1 WO2023036042 A1 WO 2023036042A1
Authority
WO
WIPO (PCT)
Prior art keywords
network optimization
network
target
service flow
strategy
Prior art date
Application number
PCT/CN2022/116441
Other languages
English (en)
French (fr)
Inventor
张欢欢
于颀
开毅
蔡明杰
俞博源
孙宸
Original Assignee
华为技术有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 华为技术有限公司 filed Critical 华为技术有限公司
Publication of WO2023036042A1 publication Critical patent/WO2023036042A1/zh

Links

Images

Classifications

    • 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
    • H04L41/142—Network analysis or design using statistical or mathematical methods
    • 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

Definitions

  • the present application relates to the communication field, and in particular to a method, device and system for determining a network optimization strategy.
  • a wide area network is a communication network that connects local area networks or metropolitan area networks in different regions.
  • WAN can usually provide the following network optimization strategies: forward error correction (forward error correction, FEC), compression, multi-send selective reception, packet-by-packet load sharing, jitter buffer, packet loss Active retransmission and transmission control protocol (transmission control protocol, TCP) acceleration, etc.
  • FEC forward error correction
  • TCP transmission control protocol
  • the operation and maintenance personnel can configure a network optimization strategy that can meet the service requirements in the network device based on the service requirements of the service flow, such as the requirements defined in the service level agreement (SLA).
  • the network device can then execute the network optimization policy on the service flow. For example, if the service requirements of a certain service flow are high quality and high throughput, the network optimization policy configured by the operation and maintenance personnel for the service flow may include FEC. If the service requirement of a service flow is high efficiency, the network optimization strategy configured by the operation and maintenance personnel for the service flow can include TCP acceleration.
  • This application provides a method, device, and system for determining a network optimization strategy, which can solve the technical problem of low efficiency in network optimization strategy configuration.
  • the technical solution is as follows:
  • a method for determining a network optimization strategy is provided, which is applied to network optimization equipment; the method includes: obtaining network performance parameters, and using a network optimization algorithm to process input parameters to obtain a target network optimization strategy, wherein the input parameters Including the performance parameters of the network and the service requirements of the target service flow; then, applying the target network optimization strategy to the target service flow, so that the transmission performance of the target service flow in the network meets the service requirements.
  • the network optimization device can automatically determine the network optimization strategy based on the network optimization algorithm, the efficiency of determining the network optimization strategy is effectively improved.
  • the network optimization device when determining the network optimization strategy, not only considers the service requirement of the service flow, but also considers the performance parameters of the network, so the determined network optimization strategy can adapt to the dynamic change of the network performance.
  • the process of acquiring the performance parameter of the network may include: acquiring the performance parameter of a link through which the target service flow will flow in the network.
  • the performance of the link that the target service flow will flow through is a key factor affecting the transmission performance of the target service flow
  • the target network optimization strategy when determining the target network optimization strategy, only the link that the target service flow will flow in the network can be obtained performance parameters. Therefore, on the premise of ensuring the accuracy of the determined target network optimization strategy, the amount of data to be processed by the network optimization device can be reduced, and the efficiency of data processing can be improved.
  • the network optimization algorithm is used to process the performance parameters of the network and the service requirements of the target service flow, and the process of obtaining the target network optimization strategy may include: using the network optimization algorithm to process the input parameters, so as to select from a plurality of different types of network optimization technologies At least one network optimization technology is selected to obtain a target network optimization strategy.
  • the plurality of different types of network optimization technologies may include at least two of the following technologies: FEC technology, compression technology, multiple transmission and selective reception technology, packet-by-packet load sharing technology, jitter buffer technology, packet loss active retransmission technology and TCP acceleration technology, etc.
  • the input parameter may also include: the first temporary transmission performance of the target service flow after the temporary network optimization technology is applied to the target service flow, and the temporary network optimization technology belongs to the plurality of different types of network optimization technologies; correspondingly, Selecting at least one network optimization technology from a plurality of different types of network optimization technologies to obtain a target network optimization strategy may include: In, at least one network optimization technology different from the temporary network optimization technology is selected to obtain a target network optimization strategy.
  • the network optimization device After the network optimization device applies the temporary network optimization technology to the target service flow, it can also adjust the network optimization technology applied to the target service flow based on the actual transmission performance of the target service flow in the network, so as to ensure that the adjusted network optimization technology (That is, the target network optimization strategy) can make the transmission performance of the target service flow meet the service requirements. In this way, adaptive adjustment of the network optimization strategy applied to the target service flow can be realized, ensuring that the transmission performance of the target service flow can always meet service requirements during the dynamic change of network performance.
  • the process of obtaining the target network optimization strategy may include: using the network optimization algorithm to process the input parameters, so as to obtain a parameter value range of a network optimization technology A set of parameter values is selected to obtain the target network optimization strategy.
  • the network optimization device can select not only the type of network optimization technology, but also the parameter value of the network optimization technology, so as to obtain the target network optimization policy finally applied to the target service flow.
  • the input parameter may also include: the second temporary transmission performance of the target service flow after applying a network optimization technique whose parameter value is the first temporary parameter value to the target service flow; correspondingly, from a network optimization
  • the process of selecting a set of parameter values within the parameter value range of the technology to obtain the target network optimization strategy may include: if the second temporary transmission performance does not meet the service requirements, within the parameter value range of a network optimization technology, the first temporary The parameter values are adjusted to obtain the target network optimization strategy.
  • the network optimization device can also adaptively adjust the parameter values of the network optimization technology applied to the target service flow. In this way, it can be ensured that the transmission performance of the target service flow can always meet the service requirements during the dynamic change of the network performance.
  • the input parameters may also include: after applying a network optimization technique whose parameter value is the second temporary parameter value to the target service flow, the third temporary transmission performance of the target service flow; correspondingly, from a network optimization technique
  • the process of selecting a group of parameter values within the parameter value range of the target network optimization strategy may include: if the third temporary transmission performance is better than the service requirement, then within the parameter value range of a network optimization technology, the second temporary parameter The value is adjusted to obtain the target network optimization strategy; wherein, the device resources required to execute a network optimization technology whose parameter value is the second temporary parameter value are more than the device resources required to execute the target network optimization strategy.
  • the network optimization device can also adjust the parameter values of the network optimization technology to ensure that the consumption of the network device by the network optimization strategy is minimized on the premise of meeting the service requirement.
  • Device resources As a result, more device resources can be released to implement network optimization strategies for other service flows, that is, the resource utilization rate of the network device can be effectively improved, and the number of service flows that the network device can carry is increased.
  • the input parameter may also include: resource usage information of the network device used to implement the target network optimization strategy; by adding the resource usage information to the input parameter, it can be ensured that the device resources of the network device are sufficient to execute the target network optimization strategy. That is, it can be ensured that the determined target network optimization strategy can be effectively executed by the network device.
  • the network optimization device uses a network optimization algorithm to process input parameters, and the process of obtaining a target network optimization strategy may include: inputting network performance parameters and service requirements into an optimization model, and obtaining one or more different types of output from the optimization model A network optimization technology; wherein, the target network optimization strategy determined by the network optimization algorithm is determined from the one or more different types of network optimization technologies.
  • the network optimization algorithm may also include an online learning algorithm, and the network optimization device may output one or more different types of network optimization techniques to the optimization model based on the online learning algorithm, and/or, the configuration parameters of the network optimization techniques The parameter value is selected to obtain the target network optimization strategy.
  • the optimization model may be obtained by training training samples, and the training samples may include: a reference network optimization strategy, network performance parameters before applying the reference network optimization strategy to the reference service flow, and The transmission performance of the reference service flow after applying the reference network optimization strategy.
  • the optimization model may be trained by the network optimization device, or may be trained by a model trainer and sent to the network optimization device. Moreover, the network optimization device or the model trainer may also periodically update the optimization model to improve the performance of the optimization model.
  • the network optimization device may be a network device for executing a target network optimization strategy.
  • the process of applying the target network optimization strategy to the target service flow may include: executing the target network optimization strategy on the target service flow.
  • the network optimization device may be a controller.
  • the process of applying the target network optimization strategy to the target service flow may include: sending the target network optimization strategy to a network device for executing the target network optimization strategy, to The network device is made to execute the target network optimization strategy for the target service flow.
  • each controller can be connected to multiple network devices, using the controller as a network optimization device can realize centralized calculation of network optimization technologies for service flows passing through different network devices. Moreover, it is convenient to maintain and update the network optimization algorithm.
  • a network optimization device in another aspect, includes at least one module, and the at least one module is configured to implement the method for determining a network optimization policy provided in the above aspect.
  • a network optimization device in yet another aspect, includes: a memory, a processor, and a computer program stored on the memory and operable on the processor.
  • the processor executes the computer program, the following is implemented: The method for determining the network optimization strategy provided by the above aspect.
  • a computer-readable storage medium where instructions are stored in the computer-readable storage medium, and the instructions are executed by a processor to implement the method for determining a network optimization strategy provided in the above aspects.
  • a computer program product including instructions is provided, and when the computer program product is run on a computer, the computer is made to execute the method for determining a network optimization strategy provided in the above aspect.
  • a traffic forwarding system includes: a controller and a network device; wherein, the controller is used to execute the method for determining a network optimization strategy as provided in the above aspect, and the network device is used to implement the method for determining the target traffic flow Execute the network optimization strategy determined by the controller.
  • the present application provides a method, device and system for determining a network optimization strategy.
  • the network optimization device can use the network optimization algorithm to process the performance parameters of the network and the service requirements of the target service flow, obtain the target network optimization strategy, and apply the target network optimization strategy to the target service flow, so that the target The transmission performance of the service flow meets the service requirement. Since the network optimization device can automatically determine the network optimization strategy based on the network optimization algorithm, the efficiency of determining the network optimization strategy is effectively improved. And because the performance parameters of the network are also considered when determining the network optimization strategy, the determined network optimization strategy can adapt to the dynamic change of network performance. Therefore, when the network performance changes dynamically, it can also ensure that the transmission performance of the target service flow meets the service requirement.
  • FIG. 1 is a schematic structural diagram of a traffic forwarding system provided by an embodiment of the present application
  • FIG. 2 is a flowchart of a method for determining a network optimization strategy provided by an embodiment of the present application
  • FIG. 3 is a schematic structural diagram of another traffic forwarding system provided by an embodiment of the present application.
  • FIG. 4 is a schematic structural diagram of another traffic forwarding system provided by an embodiment of the present application.
  • FIG. 5 is a schematic diagram of an optimization model input parameters and an output network optimization strategy provided by an embodiment of the present application
  • Fig. 6 is a schematic diagram of the working principle of an optimization model provided in the embodiment of the present application.
  • FIG. 7 is a schematic diagram of a working principle of a network optimization algorithm provided in an embodiment of the present application.
  • FIG. 8 is a schematic diagram of a state machine provided by an embodiment of the present application.
  • FIG. 9 is a schematic structural diagram of another traffic forwarding system provided by an embodiment of the present application.
  • FIG. 10 is a schematic structural diagram of another traffic forwarding system provided by an embodiment of the present application.
  • FIG. 11 is a schematic structural diagram of a network optimization device provided in an embodiment of the present application.
  • FIG. 12 is a schematic structural diagram of another network optimization device provided by an embodiment of the present application.
  • FIG. 13 is a schematic structural diagram of another network optimization device provided by an embodiment of the present application.
  • FIG. 1 is a schematic structural diagram of a traffic forwarding system provided by an embodiment of the present application.
  • the system also called a network
  • the system may include a controller 01 and multiple network devices.
  • the network devices shown in FIG. 1 include a first network device 02 and a second network device 03 .
  • a communication connection is established between the controller 01 and at least one network device (for example, the first network device 02 ), and communication connections are established between various network devices.
  • the controller 01 is used for unified management and control of various network devices connected to it, and each network device is used for forwarding packets of service flows to other network devices.
  • the operation and maintenance personnel may configure the network optimization technology for the target service flow in the controller 01 , and the controller 01 may deliver the network optimization technology to the first network device 02 .
  • the first network device 02 can further execute the network optimization technology on the target service flow, and forward the packets of the target service flow to the second network device 03 .
  • the controller 01 may be a server, or a server cluster composed of several servers, or a cloud computing service center.
  • Both the first network device 02 and the second network device 03 may be network devices with a packet forwarding function, such as a router or a switch, and may also be called forwarding devices.
  • both the first network device 02 and the second network device 03 may be access routers (access router, AR).
  • the traffic forwarding system may be a wide area network, such as a software-defined wide area network (software-defined WAN, SD-WAN).
  • a software-defined wide area network software-defined WAN, SD-WAN.
  • the business flow carried in the WAN is complex and changeable, and the configurable network optimization technologies in the WAN (such as FEC, compression, multi-send and selective reception, packet-by-packet load sharing, jitter buffering, active retransmission of lost packets, and TCP acceleration, etc. ) are many, and the configuration parameters of each network optimization technology are various.
  • the efficiency of manual configuration of wide area optimization technology by operation and maintenance personnel is low.
  • the performance of the link through which the target service flow flows may deteriorate, resulting in the situation that the pre-configured network optimization technology cannot meet the service requirements of the target service flow.
  • the embodiment of the present application provides a method for determining a network optimization strategy.
  • the network optimization device can use the network optimization algorithm to process the performance parameters of the network and the service requirements of the target service flow, so as to obtain the target network that can be applied to the target service flow.
  • Optimization Strategy may include at least one network optimization technology, or may include at least one network optimization technology and parameter values of configuration parameters of at least one network optimization technology.
  • the network optimization technology may be a wide area optimization technology. Since the method provided by the embodiment of the present application does not require operation and maintenance personnel to manually determine and configure the network optimization strategy, the efficiency of network optimization strategy determination and configuration is effectively improved. And because the performance parameters of the network are also considered when determining the network optimization strategy, the determined network optimization strategy can adapt to the dynamic change of network performance. Therefore, when the network performance changes dynamically, it can also ensure that the transmission performance of the target service flow meets the service requirement.
  • Fig. 2 is a flow chart of a method for determining a network optimization strategy provided by an embodiment of the present application, and the method may be applied to a network optimization device.
  • the network optimization device may be a controller, or a network device for forwarding service flows. As shown in Figure 2, the method includes:
  • Step 101 acquiring network performance parameters.
  • the performance parameters of the network may include at least one of the following parameters: delay, packet loss rate, jitter, and throughput.
  • the performance parameter of the network may be the performance parameter of each link in the network, or may be the performance parameter of the link through which the target service flow will flow in the network.
  • the target service flow refers to a service flow that needs to be applied with a network optimization technology, and the number of links that the target service flow will flow through may be greater than or equal to one. If the number of links that the target service flow will flow through is greater than 1, that is, the target service flow has multiple optional links (also called feasible links or reachable links) in the network, the network optimization device can obtain each Optional link performance parameters.
  • each service flow in the network can be uniquely identified by a five-tuple (or four-tuple or seven-tuple).
  • the network optimization device may determine the link through which the target service flow will flow based on the unique identifier of the target service flow.
  • the link through which the target service flow will flow may include a leased line link, an ordinary Internet (Internet) link, a mobile communication link, and the like.
  • the mobile communication link may be a fourth generation mobile communication technology (4th generation mobile communication technology, 4G) link or a 5G link.
  • the network optimization device can only obtain the links that the target service flow will flow in the network. Link performance parameters. Therefore, on the premise of ensuring the accuracy of the determined target network optimization strategy, the amount of data to be processed by the network optimization device can be reduced, and the efficiency of data processing can be improved.
  • each network device in the network is equipped with an information collection module, and the information collection module can collect performance parameters of the network through path-associated measurement or active measurement.
  • the performance parameter of the network is taken as an example of the performance parameter of the link through which the target service flow will flow
  • the network optimization device is taken as the first network device 02 as an example for illustration.
  • the link through which the target service flow will flow is the link between the first network device 02 and the second network device 03, that is, the first network device 02 is the gateway device on the sending side of the target service flow, and the second network device 03 It is the gateway device on the receiving side of the target service flow.
  • the information collection module in the first network device 02 can encapsulate marking information (such as the sending timestamp of the service message and/or the global sequence number of the service message, etc.) in the service message of the target service flow. ), and send the service message to the second network device 03.
  • marking information such as the sending timestamp of the service message and/or the global sequence number of the service message, etc.
  • the information collection module in the first network device 02 may actively send a detection message dedicated to measuring performance parameters to the second network device 03, and the detection message is encapsulated with tag information.
  • the information collection module in the second network device 02 After the information collection module in the second network device 02 receives the service message or detection message encapsulated with the label information, it can calculate the performance parameters of the link based on the label information, and feed back the calculated performance parameters to the first network device 02.
  • the second network device 02 can calculate the packet loss rate and throughput of the link based on the global sequence number in the message, and can calculate the delay of the link based on the sending timestamp in the message.
  • the second network device 02 may feed back the receiving timestamp of the packet to the first network device 02, and the first network device 02 calculates the link delay according to the sending timestamp and receiving timestamp of the packet.
  • the first network device 01 may send a service packet or a probe packet encapsulated with tag information to the second network device 02 .
  • the information collection module in the first network device 01 may send the detection information of the message to the information receiving module of the controller 01, such as the tag information encapsulated in the message, and/or, the tag information encapsulated The number of message packets.
  • the information collection module in the second network device 02 After the information collection module in the second network device 02 receives the message encapsulated with the tag information, it can also send the detection information of the received message to the information receiving module of the controller 01, such as the receiving time stamp of the message, and /or, the number of received packets encapsulated with tag information.
  • the information receiving module of the controller 01 can further calculate the performance parameter of the link based on the detection information sent by the first network device 01 and the second network device 02 .
  • Step 102 acquiring resource usage information of the network device.
  • the network device at least includes a first network device for performing network optimization technology on the target service flow, that is, a gateway device on the sending side of the target service flow.
  • the network device may also include a gateway device (that is, a second network device) on the receiving side of the target service flow in the network.
  • the resource usage information of the network device may include at least one of the following information: number of sessions, processor usage information, memory usage information, and the like.
  • the processor may include one or more of a central processing unit (central processing unit, CPU), a network processing unit (network processing unit, NPU), and a graphics processing unit (graphics processing unit, GPU). It can be understood that the resource usage information may be represented by the amount of resources used by the network device, the percentage of used resources, and/or the amount of remaining available resources.
  • the information collection module of the first network device 02 may collect its own resource usage information.
  • the information collection module of the second network device 03 may send the resource usage information of the second network device 03 to the first network device 02 .
  • the information collection module of the first network device 02 may send the resource usage information of the first network device 02 to the information receiving module of the controller 01 .
  • the information collecting module of the second network device 03 may send the resource usage information of the second network device 03 to the information receiving module of the controller 01 .
  • Step 103 using a network optimization algorithm to process input parameters to obtain a target network optimization strategy, where the input parameters include network performance parameters, resource usage information, and service requirements of target service flows.
  • the network optimization algorithm and the service requirements of the target service flows requiring network optimization are pre-configured in the network optimization device.
  • the business requirements also referred to as performance requirements or service quality requirements
  • the business requirements may be SLA requirements or other agreed requirements
  • the business requirements may include requirements for at least one of the following performance parameters: delay, packet loss rate, Jitter and throughput etc.
  • Network optimization equipment can periodically obtain network performance parameters and resource usage information of network devices, and can periodically use network optimization algorithms to process the performance parameters, resource usage information and business requirements of target business flows to obtain target network optimization strategies .
  • the network optimization device may select at least one network optimization technology from a plurality of different types of network optimization technologies, and/or select a network optimization technology within the parameter value range of a network optimization technology. Group parameter values, so as to obtain the target network optimization strategy.
  • each network optimization technique has at least one configuration parameter, and the parameter value range of each network optimization technique may include an optional value range of each configuration parameter of the network optimization technique.
  • a set of parameter values of each network optimization technology may include a parameter value of at least one configuration parameter of this network optimization technology.
  • the target network optimization strategy may include at least one network optimization technology, or the target network optimization strategy may include at least one network optimization technology, and in the at least one network optimization technology, one or more Parameter values for configuration parameters for network optimization techniques.
  • the multiple different types of network optimization technologies may include FEC technology, compression technology, multi-send and selective reception technology, packet-by-packet load sharing technology, jitter buffer technology, packet loss active retransmission technology, TCP acceleration technology and seven-layer protocol optimization techniques, etc.
  • the seven-layer protocol optimization technology may include: a hypertext transfer protocol (hyper text transfer protocol, HTTP) optimization technology, a file transfer protocol (file transfer protocol, FTP) optimization technology, and the like.
  • the configuration parameters of FEC technology can include encoding block size (block size), packet loss rate and redundancy rate, etc.;
  • the configuration parameters of compression technology can include the type of compression algorithm and the size of sliding window, etc.;
  • the parameters can include reassembly waiting time and jitter, etc.;
  • the configuration parameters of the packet-by-packet load sharing technology can include the weight of each link and the reassembly waiting time;
  • the configuration parameters of the jitter buffer technology can include cache size and cache waiting time; packet loss initiative
  • the configuration parameters of the retransmission technology can include: cache size, cache waiting time, and message release waiting time at the sending end, etc.;
  • the configuration parameters of the TCP acceleration technology can include: the size of the congestion window and the slow start threshold, etc.;
  • the configuration of the HTTP optimization technology Parameters can include resource caching strategies and data prefetch methods, etc.
  • the encoding block size refers to the number of messages required to be separated between two adjacent redundant packet encoding operations. For example, assuming that the encoding block size is 20, it means that a redundant packet encoding operation is performed every 20 packets, and each redundant packet encoding operation can generate one or more redundant packets.
  • the resource caching strategy in the HTTP optimization technology may include a least recently used (least recently used, LRU) strategy and a least recently used (least frequently used, LFU) strategy.
  • At least one target link may also be determined from the multiple optional links.
  • the at least one target link is a link for forwarding service packets of the target service flow.
  • the network optimization device may not perform the above step 102, and correspondingly, the input parameter may not include resource usage information of the network device.
  • Step 104 Apply a target network optimization policy to the target service flow, so that the transmission performance of the target service flow in the network meets the service requirement.
  • the network optimization device After the network optimization device determines the target network optimization strategy, it can apply the target network optimization strategy to the target service flow. Since the network optimization device considers the performance parameters of the network when determining the target network optimization strategy, even if the network performance changes dynamically, it can be ensured that after the target network optimization strategy is applied, the transmission performance of the target service flow in the network can still meet the requirements. The business requirements of the target business flow. If the input parameter also includes resource usage information of the network device, it can be ensured that the network device used to execute the target network optimization strategy has sufficient device resources to execute the target network optimization strategy. That is, it can be ensured that the determined target network optimization strategy can be effectively executed by the network device.
  • the first network device 02 is configured with an optimization module and an execution module.
  • the optimization module uses a network optimization algorithm to process input parameters, and after obtaining the target network optimization strategy, the execution module can execute the target network optimization strategy for the target service flow. That is, the execution module can process the service packets of the target service flow by adopting the target network optimization strategy.
  • the controller 01 is configured with an optimization module.
  • the optimization module uses a network optimization algorithm to process input parameters, and after obtaining a target network optimization strategy, it can deliver the target network optimization strategy to the first network device 02 .
  • the execution module of the first network device 02 can further execute the target network optimization policy on the target service flow.
  • controller can be connected to multiple network devices, using the controller as a network optimization device can realize centralized calculation of network optimization technologies for service flows passing through different network devices. Moreover, it is convenient to maintain and update the network optimization algorithm.
  • the first network device 02 may further include a traffic identification module, an encapsulation module and a forwarding module.
  • the traffic identification module is used to identify the target service flow requiring network optimization, for example, the traffic identification module can identify the target service flow based on the five-tuple of the service flow.
  • the encapsulation module is used to encapsulate the service message of the target service flow.
  • the encapsulation module can use generic routing encapsulation (GRE) to encapsulate the identification of the target network optimization strategy in the service message.
  • GRE generic routing encapsulation
  • the forwarding module is configured to forward the service packet of the target service flow to the second network device 03, for example, the forwarding module may forward the service packet of the target service flow through at least one target link.
  • the second network device 03 further includes a decapsulation module, a logic processing module and a forwarding module.
  • the decapsulation module is used for decapsulating the received service message.
  • the logical processing module is used to perform logical processing on the service message, for example, based on the identification of the target network optimization strategy encapsulated in the service message, the service message may be decoded by using a corresponding decoding technology.
  • the forwarding module is used for forwarding logically processed service packets to the receiver device of the target service flow.
  • the network optimization device may continue to execute the foregoing steps 101 to 103, so as to update the target network optimization policy.
  • the network optimization device may periodically perform the above steps 101 to 103; or, the network optimization device may perform the above steps 101 to 103 again when it detects that the performance parameters of the network and/or the resource usage information of the network device change.
  • Step 103 the network optimization device can adaptively adjust the network optimization strategy applied to the target service flow, so as to ensure that the service requirements of the target service flow can be met.
  • the network optimization algorithm deployed in the network optimization device includes an optimization model (also called an adaptive model).
  • the network optimization device can input the obtained network performance parameters, resource usage information of network devices, and service requirements of the target service flow into the optimization model, and obtain one or more different types of network optimization technologies output by the optimization model.
  • the network optimization device can determine a target network optimization strategy from one or more different types of network optimization techniques output by the optimization model. For example, the network optimization device may randomly select a network optimization technology to obtain a target network optimization policy.
  • the optimization model may output at least one network optimization strategy (policy), each network optimization strategy includes: at least one target link identification, at least one network optimization technology, and each network optimization technology The parameter value for the configuration parameter.
  • policy network optimization strategy
  • each network optimization strategy includes: at least one target link identification, at least one network optimization technology, and each network optimization technology The parameter value for the configuration parameter.
  • the optimization model can output n network optimization strategies, where n is an integer greater than 1.
  • the identifier of the target link in the network optimization policy P1 is: link 1, the network optimization technology is FEC technology, and the parameter values of the FEC technology include: the coding block size is 10, and the packet loss rate is 2%.
  • the network optimization technology in the network optimization policy P2 includes packet-by-packet load sharing technology, and the identification of the target link includes: link 1 and link 2, that is, the link 1 and link 2 are used to implement packet-by-packet load sharing.
  • the network optimization technology in the network optimization policy P2 further includes: FEC technologies corresponding to link 1 and link 2 respectively.
  • the parameter values of the FEC technology corresponding to link 1 include: the coding block size is 30, and the packet loss rate is 1%; the parameter values of the FEC technology corresponding to link 2 include: the coding block size is 20, and the packet loss rate is 1%. 10%.
  • the identification of the target link in the network optimization strategy Pn is: link K
  • the network optimization technology is the multi-send and selective reception technology
  • the parameter values of the multi-send and selective reception technology include: the reassembly waiting time is 100 milliseconds (ms), and the jitter is 10 ms .
  • the optimization model may be a rule-based model, and for different types of service flows, the content contained in the network optimization policy output by the optimization model may be different.
  • the optimization model can be based on the business requirements of the service flow for the service flow Select the target link, and at least one network optimization technique.
  • the optimization model may determine parameter values of configuration parameters of the network optimization technology based on network performance parameters (or performance parameters and resource usage information of network devices). For example, for the FEC technology, the size of the coding block and the packet loss rate can be determined according to the performance parameters of the link.
  • the network optimization policy output by the optimization model may include: an identification of at least one target link, at least one network optimization technology, and parameter values of configuration parameters of each network optimization technology.
  • the optimization model can select a target link for the service flow only based on service requirements, without needing to determine a network optimization technology.
  • the network optimization policy output by the optimization model may only include the identification of at least one target link.
  • the network optimization strategy for the target type of service flow output by the optimization model may also include only the identification of at least one target link, without including the network optimization technology.
  • the network optimization device does not need to apply the network optimization technology to the service flow of the target type. Therefore, it can be ensured that when the network performance becomes better, the network optimization device can timely disable the network optimization technology for the target service flow, thereby effectively releasing device resources of the network device.
  • the optimization model determines the rules of the network optimization technology based on the input parameters, which may be determined based on the characteristics of different network optimization technologies combined with experimental analysis.
  • Table 1 schematically shows the characteristics of different network optimization technologies.
  • FEC technology is a network optimization technology that uses redundant packets to resist packet loss and reduce retransmission delay.
  • the throughput can also be improved in the scene. For example, the throughput of a link with a large bandwidth-delay product (BDP) packet loss or a link with a high non-blocking packet loss rate can be improved.
  • BDP bandwidth-delay product
  • the usage constraints of this FEC technology are mainly: introducing additional bandwidth to resist packet loss, and not applicable to blocked links. Wherein, non-blocking packet loss is also referred to as random packet loss, which refers to packet loss not caused by link congestion.
  • the main purpose of TCP acceleration technology, packet-by-packet load sharing technology, and compression technology is to improve throughput
  • the main purpose of multi-send and selective reception technology is to achieve high reliability.
  • different network optimization technologies may have different restrictions on the number of sessions. If a network optimization technology limits the number of sessions to M (M is an integer greater than 1), it means that the network device can execute the network optimization technology on M service flows at most at the same time. For example, the limit M of the number of sessions imposed by the FEC technology may be 32.
  • the network optimization algorithm may also include an online learning algorithm.
  • the online learning algorithm can select at least one network optimization technology from multiple different types of network optimization technologies output by the optimization model based on the transmission performance of the target service flow in the network to obtain the target network optimization strategy, and/or, from the optimization A set of parameter values is selected within the parameter value range of a network optimization technique output by the model to obtain the target network optimization strategy. That is, the network optimization device can select the type of network optimization technology output by the optimization model and/or the parameter value of the network optimization technology through an online learning algorithm to obtain a final target network optimization strategy.
  • the transmission performance of the target service flow in the network can be represented by at least one of the following performance parameters: time delay, packet loss rate, jitter, and throughput.
  • the input parameters processed by the network optimization algorithm may also include: after applying the temporary network optimization technique to the target service flow, The first temporary transmission performance of the target service flow.
  • the temporary network optimization technology belongs to the plurality of different types of network optimization technologies; correspondingly, the process for the network optimization device to determine the target network optimization strategy through an online learning algorithm may include:
  • the optimization model selects at least one network optimization technology different from the temporary network optimization technology to obtain the target Network optimization strategy.
  • the network optimization device may select another network optimization technique different from the temporary network optimization technique to obtain a target network optimization strategy.
  • the network optimization device may select another network optimization technique different from the temporary network optimization technique, and combine the temporary network optimization technique with the other network optimization technique to obtain a target network optimization strategy. That is, the network optimization device can apply multiple network optimization technologies to the target service flow.
  • the network optimization technology output by the optimization model includes the FEC technology and the multiple transmission and selective reception technology. If the network optimization device applies the FEC technology as a temporary network optimization technology to the target service flow, and the first temporary transmission performance of the target service flow in the network does not meet the service requirements, the network optimization device can determine the multi-send and selective reception technology as Target network optimization strategy, and apply the multi-send and select-receive technology to the target traffic flow. Alternatively, the network optimization device may determine the combination of the FEC technology and the multi-feed and selective reception technology as a target network optimization strategy, and apply the FEC technology and the multi-feed and selective reception technology to the target service flow respectively.
  • the network optimization device after the network optimization device applies the temporary network optimization technology to the target service flow, it can also apply the corresponding The network optimization technology of the service flow is adjusted, so it can be ensured that the adjusted network optimization technology (that is, the target network optimization strategy) can make the transmission performance of the target service flow meet the service requirement.
  • the adjusted network optimization technology that is, the target network optimization strategy
  • adaptive adjustment of the network optimization strategy applied to the target service flow can be realized, ensuring that the transmission performance of the target service flow can always meet service requirements during the dynamic change of network performance.
  • the input parameters processed by the network optimization algorithm may also include: applying a method whose parameter value is the first temporary parameter value to the target service flow After the network optimization technology, the second temporary transmission performance of the target service flow.
  • the process for the network optimization device to determine the target network optimization strategy through the online learning algorithm may include:
  • the first temporary parameter value is adjusted to obtain the target network optimization strategy.
  • the parameter value range of the network optimization technology may be output by the optimization model, and the parameter value range may include an optional value range of each configuration parameter of the network optimization technology.
  • the network optimization device may adjust the parameter value of at least one configuration parameter within the optional value range of the parameter value until the transmission of the target service flow Performance meets business needs.
  • the parameter value range of the FEC technology output by the optimization model may include: Select the value range, and the optional value range of the packet loss rate.
  • the network optimization device may adjust the coding block size and/or the packet loss rate until the transmission performance of the target service flow meets the service requirements.
  • the network optimization device when the network optimization device adjusts the parameter value of the configuration parameter, it may first adjust the parameter value of the configuration parameter in the direction of increasing the parameter value and decreasing the parameter value, and respectively obtain the target service The first updated transmission performance of the target service flow after the network optimization technology with increased parameter value is applied to the flow, and the second updated transmission performance of the target service flow after the network optimization technology with reduced parameter value is applied to the target service flow. Afterwards, the network optimization device may use a benefit function (such as the Mahalanobis distance) to respectively determine the first distance between the first updated transmission performance and the service requirement of the target service flow, and the distance between the second updated transmission performance and the service requirement of the target service flow. second distance.
  • a benefit function such as the Mahalanobis distance
  • the network optimization device can continue to increase the parameter value within the optional value range of the configuration parameter until the target service flow After applying the network optimization technology that increases the parameter value, the transmission performance of the target service flow meets the service requirements. If the second distance is smaller than the first distance, the network optimization device can continue to reduce the parameter value within the optional value range of the configuration parameter until the target service flow is applied to the network optimization technology that reduces the parameter value, and the target service flow The transmission performance meets business requirements.
  • the network optimization device may first adjust the parameter value of the configuration parameter according to the target direction, and obtain the network optimization technology for applying the adjusted parameter value to the target service flow. , the update transmission performance of the target traffic flow. If the updated transmission performance is closer to the service requirement of the target service flow than the second temporary transmission performance, the network optimization device may continue to adjust the parameter value according to the target direction until the network optimization technology for adjusting the parameter value is applied to the target service flow , the transmission performance of the target service flow meets the service requirements.
  • the network optimization device may adjust the parameter value in a direction opposite to the target direction until the network optimization for adjusting the parameter value is applied to the target service flow After the technology, the transmission performance of the target service flow meets the service requirements.
  • the target direction may be a direction of increasing a parameter value or a direction of decreasing a parameter value.
  • the network optimization device may adjust the parameter value of the configuration parameter according to a fixed adjustment ratio. For example, the network optimization device may increase the parameter value to 1.25 times the original parameter value each time, or decrease it to 0.75 times the original parameter value.
  • the network optimization device can re- The network optimization algorithm is used to process the input parameters to obtain the target network optimization strategy.
  • the network optimization device may determine the target network optimization strategy in combination with the above first possible implementation manner and the second possible implementation manner. For example, if the network optimization device determines that the first temporary transmission performance of the target service flow does not meet the service requirements, it may first adjust the parameter value of the configuration parameter of the temporary network optimization technology in the second possible implementation manner above. If the parameter value of the configuration parameter of the temporary network optimization technology is adjusted to the upper limit or lower limit of the optional value range, but still cannot meet the service requirements, the network optimization device can adopt the first possible implementation method above, and select the temporary At least one network optimization technique different from the network optimization technique obtains a target network optimization strategy.
  • the network optimization device may first use the network optimization technology P1 as a temporary network optimization technology and apply it to the target service flow. Afterwards, the network optimization device may detect the first temporary transmission performance U1 of the target service flow in the network. If the first temporary transmission performance U1 does not meet the service requirement T of the target service flow, the network optimization device may adjust the parameter value of the configuration parameter of the network optimization technology P1 from A 1 to A 2 and apply it to the target service flow. Then, the network optimization device may detect the second temporary transmission performance U2 of the target service flow in the network. If the second temporary transmission performance U2 still does not meet the service requirement T, the network optimization device may continue to adjust the parameter values of the configuration parameters of the network optimization technology P1.
  • the network optimization device can apply the network optimization technology to the target service flow P2. If the transmission performance of the target service flow meets the service requirement T after applying the network optimization technology P2, the network optimization device can use the network optimization technology P2 as a target network optimization strategy and continue to apply it to the target service flow.
  • i in FIG. 7 is a positive integer
  • k is an integer greater than 1
  • Ui represents the transmission performance of the target service flow obtained by the network optimization device for the ith time.
  • the input parameter may further include: after applying a network optimization technology whose parameter value is the second temporary parameter value to the target service flow, the third Temporary transmission performance; the process in which the network optimization device determines the target network optimization strategy through an online learning algorithm may include:
  • the second temporary parameter value is adjusted within the parameter value range of the network optimization technology to obtain the target network optimization strategy.
  • the device resources required to execute the network optimization technology whose parameter value is the second temporary parameter value are more than the device resources required to execute the target network optimization strategy.
  • the network optimization device can also adjust the second temporary parameter value of the network optimization technology, so as to ensure that the network optimization strategy is minimized while meeting the business requirement.
  • Device resources consumed by network devices that is, the resource utilization of network devices can be effectively improved, and the number of service flows that can be carried by the network device (that is, the number of supported sessions) can be increased. ).
  • the network optimization device may adjust the value of the second temporary parameter according to the two optional examples described above, so that after the network optimization technology with the adjusted parameter value is applied to the target service flow, the target service The transmission performance of the stream meets the business requirement and is closer to the business requirement.
  • a state machine is also configured in the network optimization device.
  • the network optimization device can adjust the parameter value of the network optimization technology or switch the network optimization strategy according to the state machine.
  • Types of technology The following takes the state machine shown in FIG. 8 as an example to introduce the process of the network optimization device determining the target network optimization policy.
  • the network optimization device in the initialization (startup) state, can determine an initial network optimization technique (ie, a temporary network optimization technique) based on the output of the optimization model.
  • the network optimization device may apply the temporary network optimization technology to the target service flow.
  • the network optimization device enters into a continuous optimization (continue optimization) state, and executes step S2 every adjustment interval, wherein the adjustment interval may be 1 second.
  • the network optimization device may adjust the parameter value of the configuration parameter of the temporary network optimization technology, and detect whether the temporary transmission performance of the target service flow meets the service requirement after the parameter value is adjusted.
  • step S3 if the network optimization device detects that the previous state (pre-state) is an initialization state, and the temporary transmission performance of the current target service flow does not meet service requirements, step S4 may be performed.
  • the network optimization device may select at least one network optimization technique different from the temporary network optimization technique.
  • step S5 if the network optimization device detects that the previous state is the continuous optimization state, and the temporary transmission performance of the current target service flow does not meet the service requirements, it can enter the rollback state and execute step S6.
  • the network optimization device may restore the parameter value of the configuration parameter of the temporary network optimization technology to the value before the last adjustment. That is to say, when the network optimization device continuously adjusts the parameter values, if after several previous adjustments, the temporary transmission performance of the target service flow can meet the service requirements, but after the last adjustment, the temporary transmission performance of the target service flow is not enough. Then meet business needs.
  • the network optimization device can determine that: before the last adjustment, the temporary transmission performance of the target service flow has approached the service requirement, so the parameter value of the configuration parameter of the temporary network optimization technology can be rolled back to the value before the last adjustment, so as to obtain the target Network optimization strategy.
  • the network optimization device may enter a stable (stable) state, and execute step S7.
  • the network optimization device may perform a continuous optimization operation every checking interval, that is, adjust the parameter value every checking interval, and check whether the transmission performance of the target service flow meets the service requirement after adjusting the parameter value.
  • the check interval is longer than the adjustment interval, for example, the check interval may be 5 seconds or 10 seconds.
  • the network optimization device may perform step S8.
  • the network optimization device may roll back the parameter value of the configuration parameter of the currently selected network optimization technology to the value before the last adjustment, and may increase the checking interval. For example, the check interval can be doubled. Afterwards, the network optimization device can perform continuous optimization operations according to the increased detection interval.
  • the network optimization device in a stable state, can enter a reset (reset) if it detects that the target link through which the target service flow flows is faulty (down) through step S9, or the performance of the target link continues to deteriorate. ) state, and execute step S11.
  • the network optimization device may reselect at least one network optimization technique different from the temporary network optimization technique. It can be seen from FIG. 8 that, in the continuous optimization state, the network optimization device may also enter the reset state if it detects that the target link through which the target service flow flows is faulty or the performance continues to deteriorate through step S10.
  • the optimization model may be obtained by training a plurality of training samples.
  • it may be obtained by using a machine learning algorithm (such as a decision tree or a Bayesian network, etc.) or a reinforcement learning algorithm.
  • each training sample may include: a reference network optimization strategy, performance parameters of the network before applying the reference network optimization strategy to the reference traffic flow, and the reference traffic flow after applying the reference network optimization strategy to the reference traffic flow transmission performance.
  • the reference network optimization policy may include at least one network optimization technology, or may include at least one network optimization technology, and parameter values of one or more network optimization technologies in the at least one network optimization technology.
  • the at least one network optimization technology may include FEC technology, compression technology, multiple transmission and selective reception technology, packet-by-packet load sharing technology, jitter buffer technology, packet loss active retransmission technology and/or TCP acceleration technology, etc.
  • the reference service flows in different training samples may be service flows of different types of applications, for example, may be service flows of voice services or service flows of video services.
  • the traffic forwarding system may also include a model trainer 04, which may be a server, or a server cluster composed of several servers, or a cloud computing Service Center.
  • the model trainer 04 can be used to train multiple training samples to obtain an optimized model, and deliver the optimized model to the network optimization device.
  • the model trainer 04 may deliver the optimization model to the optimization module in the first network device 02 .
  • the model trainer 04 can deliver the optimization model to the optimization module in the controller 01 .
  • each training sample used for training the model may be collected by the information collection module in each network device and reported to the model trainer 04, or may be collected by the controller 01 and reported to the model Trainer 04.
  • model trainer 04 can also periodically acquire training samples, and update the optimized model based on the acquired training samples. Afterwards, the model trainer 04 may deliver the updated optimization model to the network optimization device, and the network optimization device may then use the updated optimization model to determine a network optimization strategy.
  • model trainer 04 can also be integrated in the network optimization device, that is, the network optimization device also has the functions of model training and model update.
  • the above network optimization algorithm may also be a network optimization model trained by using a machine learning algorithm or a deep learning algorithm. That is, after the network optimization device inputs the input parameters into the network optimization model, it can obtain the target network optimization strategy output by the network optimization model.
  • step 102 may be performed before step 101, or the network optimization device may not need to perform step 102.
  • the embodiment of the present application provides a method for determining a network optimization strategy.
  • the network optimization device can use the network optimization algorithm to process the performance parameters of the network and the service requirements of the target service flow, obtain the target network optimization strategy, and apply the target network optimization strategy to the target service flow, so that the transmission performance of the target service flow meets the requirements of the target service flow.
  • the network optimization device can adjust the parameter value of the network optimization technology, switch the type of the network optimization technology, and/or superimpose other network optimization technologies to achieve The transmission performance of the target service flow is improved to ensure that the service requirements of the target service flow are met.
  • the network optimization device can release the resources of the network device by adjusting the parameter value of the network optimization technology, or disabling the network optimization technology, so as to improve the resource utilization rate of the network device.
  • the embodiment of the present application also provides a network optimization device.
  • the network optimization device can be applied to the system shown in FIG. 1, FIG. 3, FIG. 4, FIG. 9 or FIG. Methods.
  • the network optimization device 30 includes:
  • An acquisition module 301 configured to acquire network performance parameters.
  • the acquisition module 301 For the function implementation of the acquisition module 301, reference may be made to the relevant description of step 101 in the above method embodiment.
  • the optimization module 302 is configured to use a network optimization algorithm to process input parameters to obtain a target network optimization strategy.
  • the input parameters include network performance parameters and service requirements of target service flows.
  • the optimization module 302 reference may be made to the relevant description of step 103 in the above method embodiment.
  • the application module 303 is configured to apply a target network optimization strategy to the target service flow, so that the transmission performance of the target service flow in the network meets service requirements.
  • a target network optimization strategy to the target service flow, so that the transmission performance of the target service flow in the network meets service requirements.
  • the obtaining module 301 is configured to obtain performance parameters of links that the target service flow will pass through in the network.
  • the optimization module 302 may be configured to select at least one network optimization technology from a plurality of different types of network optimization technologies to obtain a target network optimization strategy.
  • the input parameters also include: the first temporary transmission performance of the target service flow after the temporary network optimization technology is applied to the target service flow, and the temporary network optimization technology belongs to multiple different types of network optimization technologies; the optimization module 302 can It is used to select at least one network optimization technology different from the temporary network optimization technology from a plurality of different types of network optimization technologies to obtain a target network optimization strategy if the first temporary transmission performance does not meet the service requirement.
  • the optimization module 302 may be configured to use a network optimization algorithm to process input parameters, so as to select a set of parameter values from a parameter value range of a network optimization technique to obtain a target network optimization strategy.
  • the input parameter also includes: after applying a network optimization technology with a parameter value of the first temporary parameter value to the target service flow, the second temporary transmission performance of the target service flow; the optimization module 302 can be used if If the second temporary transmission performance does not meet the service requirements, the first temporary parameter value is adjusted within the parameter value range of a network optimization technology to obtain a target network optimization strategy.
  • the input parameter also includes: the third temporary transmission performance of the target service flow after applying a network optimization technology whose parameter value is the second temporary parameter value to the target service flow; the optimization module 302 can be used if The third temporary transmission performance is better than the service requirement, then within the parameter value range of a network optimization technology, the second temporary parameter value is adjusted to obtain the target network optimization strategy;
  • the device resources required to execute a network optimization technique whose parameter value is the second temporary parameter value are more than the device resources required to execute the target network optimization strategy.
  • the input parameter further includes: resource usage information of the network device used to execute the target network optimization strategy; the resource usage information is used to make the device resources of the network device sufficient to execute the target network optimization strategy.
  • resource usage information is used to make the device resources of the network device sufficient to execute the target network optimization strategy.
  • the optimization module 302 can be used to input network performance parameters and business requirements into the optimization model, and obtain one or more different types of network optimization technologies output by the optimization model;
  • the target network optimization strategy is determined from the one or more different types of network optimization techniques.
  • the optimization model is obtained by training training samples, and the training samples include: a reference network optimization strategy, network performance parameters before applying the reference network optimization strategy to the reference traffic flow, and reference network optimization strategy to the reference traffic flow After optimizing the policy, refer to the transmission performance of the service flow.
  • the network optimization device 30 may be a controller, and the application module 303 may be configured to issue a target network optimization policy to a network device for executing the target network optimization policy.
  • the obtaining module 301 may be the information receiving module in the foregoing embodiments.
  • the network optimization device 30 may be a network device for executing the target network optimization strategy, and the application module 303 may be used for executing the target network optimization strategy for the target service flow.
  • the acquisition module 301 may be the information acquisition module in the above embodiments.
  • the embodiment of the present application provides a network optimization device
  • the network optimization device can use the network optimization algorithm to process the performance parameters of the network and the service requirements of the target service flow, obtain the target network optimization strategy, and can optimize the target service
  • the target network optimization policy is applied to the flow, so that the transmission performance of the target service flow meets the service requirement. Since the network optimization device can automatically determine the network optimization strategy based on the network optimization algorithm, the efficiency of determining the network optimization strategy is effectively improved. And because the performance parameters of the network are also considered when determining the network optimization strategy, the determined network optimization strategy can adapt to the dynamic change of network performance. Therefore, when the network performance changes dynamically, it can also ensure that the transmission performance of the target service flow meets the service requirement.
  • the network optimization device provided in the embodiment of the present application may be implemented by an application-specific integrated circuit (ASIC), or by a programmable logic device (PLD), and the above-mentioned PLD may be a complex program Logic device (complex programmable logical device, CPLD), field-programmable gate array (field-programmable gate array, FPGA), general array logic (generic array logic, GAL) or any combination thereof.
  • ASIC application-specific integrated circuit
  • PLD programmable logic device
  • the above-mentioned PLD may be a complex program Logic device (complex programmable logical device, CPLD), field-programmable gate array (field-programmable gate array, FPGA), general array logic (generic array logic, GAL) or any combination thereof.
  • the method for determining the network optimization strategy provided by the above method embodiment may also be implemented by software.
  • each module in the above network optimization device may also be a software module.
  • Fig. 12 is a schematic structural diagram of another network optimization device provided by an embodiment of the present application.
  • the network optimization device can be applied to the systems shown in Fig. 1, Fig. 3, Fig. 4, Fig. 9 or Fig. 10, and can be used for The methods provided in the foregoing method embodiments are realized.
  • the network optimization device may include: a processor 401 , a memory 402 , a network interface 403 and a bus 404 .
  • the bus 404 is used to connect the processor 401 , the memory 402 and the network interface 403 .
  • the communication connection with other devices can be realized through the network interface 403 (which may be wired or wireless).
  • a computer program 4021 is stored in the memory 402, and the computer program 4021 is used to realize various application functions.
  • the processor 401 may be a CPU, and the processor 401 may also be other general-purpose processors, digital signal processing (digital signal processing, DSP), application-specific integrated circuit (application-specific integrated circuit) , ASIC), field-programmable gate array (field-programmable gate array, FPGA), general array logic (generic array logic, GAL) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc., or the above Any combination of processors.
  • DSP digital signal processing
  • ASIC application-specific integrated circuit
  • FPGA field-programmable gate array
  • GAL general array logic
  • a general purpose processor may be a microprocessor or any conventional processor or the like.
  • Memory 402 may include volatile memory, non-volatile memory, or a combination thereof.
  • the non-volatile memory can include read-only memory (read-only memory, ROM), programmable read-only memory (programmable ROM, PROM), erasable programmable read-only memory (erasable PROM, EPROM), electrically programmable Erases programmable read-only memory (electrically EPROM, EEPROM), flash memory, or any combination thereof.
  • the volatile memory may include random access memory (random access memory, RAM), such as static random access memory (static RAM, SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (synchronous DRAM, SDRAM), double data rate synchronous dynamic random access memory (double data date SDRAM, DDR SDRAM), enhanced synchronous dynamic random access memory (enhanced SDRAM, ESDRAM), synchronous connection dynamic random access memory (synchlink DRAM, SLDRAM ), direct memory bus random access memory (direct rambus RAM, DR RAM), or any combination thereof.
  • RAM random access memory
  • RAM random access memory
  • static random access memory static random access memory
  • DRAM dynamic random access memory
  • SDRAM synchronous dynamic random access memory
  • double data rate synchronous dynamic random access memory double data date SDRAM, DDR SDRAM
  • enhanced synchronous dynamic random access memory enhanced synchronous dynamic random access memory
  • direct memory bus random access memory direct rambus RAM, DR RAM
  • bus 404 may also include a power bus, a control bus, a status signal bus, and the like. However, for clarity of illustration, the various buses are labeled as bus 404 in the figure.
  • the processor 401 may be configured to execute the computer program 4021 stored in the memory 402 to implement the method for determining a network optimization policy provided in the foregoing method embodiments.
  • Fig. 13 is a schematic structural diagram of another network optimization device provided by an embodiment of the present application.
  • the network optimization device can be applied to the systems shown in Fig. 1, Fig. 3, Fig. 4, Fig. 9 or Fig. 10, and can be used for The methods provided in the foregoing method embodiments are implemented.
  • the network optimization device includes a main control board 501 and an interface board 502 .
  • the main control board 501 is also called a main processing unit (main processing unit, MPU) or a route processing card (route processor card). Management, device maintenance, and protocol processing functions.
  • the main control board 501 includes: a central processing unit 5011 and a memory 5012 .
  • the interface board 502 is also called a line interface unit card (line processing unit, LPU), a line card (line card), or a service board.
  • the interface board 502 is used to provide various service interfaces and realize forwarding of service frames.
  • the service interface includes but is not limited to an Ethernet interface, a SONET/SDH-based service frame (packet over SONET/SDH, POS) interface, and the like.
  • SONET refers to synchronous optical network
  • SDH refers to synchronous digital hierarchy.
  • the Ethernet interface is, for example, a flexible Ethernet service interface (flexible Ethernet clients, FlexE Clients).
  • the interface board 502 includes: a central processing unit 5021 , a network processor 5022 , a forwarding entry storage 5024 and a physical interface card (physical interface card, PIC) 5023 .
  • the CPU 5021 on the interface board 502 is used to control and manage the interface board 502 and communicate with the CPU 5011 on the main control board 501 .
  • the network processor 5022 is configured to implement message forwarding processing.
  • the form of the network processor 5022 may be a forwarding chip.
  • the forwarding chip may be a network processor (network processor, NP).
  • the forwarding chip can be realized by ASIC or FPGA.
  • the network processor 5022 is configured to forward the received message based on the forwarding table stored in the forwarding table item memory 5024, and if the destination address of the message is the address of the network optimization device, then send the message to the CPU (such as central processing unit 5021) processing; If the destination address of the message is not the address of the network optimization device, then according to the destination address, find out the next hop and the outgoing interface corresponding to the destination address from the forwarding table, and forward the message to The outbound interface corresponding to the destination address.
  • the processing of the uplink message may include: processing of the inbound interface of the message, forwarding table search; the processing of the downlink message may include: forwarding table search and so on.
  • the central processing unit can also perform the function of the forwarding chip, such as implementing software forwarding based on a general-purpose CPU, so that no forwarding chip is needed in the interface board.
  • the physical interface card 5023 is used to implement the interconnection function of the physical layer, through which the original flow enters the interface board 502 , and the processed packets are sent out from the physical interface card 5023 .
  • the physical interface card 5023 is also called a daughter card, which can be installed on the interface board 502, and is responsible for converting the photoelectric signal into a message, checking the validity of the message and forwarding it to the network processor 5022 for processing.
  • the central processor 5021 can also execute the functions of the network processor 5022 , such as implementing software forwarding based on a general-purpose CPU, so that the physical interface card 5023 does not need the network processor 5022 .
  • the network optimization device includes multiple interface boards.
  • the network optimization device further includes an interface board 503
  • the interface board 503 includes: a central processing unit 5031 , a network processor 5032 , a forwarding entry storage 5034 and a physical interface card 5033 .
  • the functions and implementation methods of the components in the interface board 503 are the same as or similar to those of the interface board 502 , and will not be repeated here.
  • the network optimization device further includes a switching fabric unit 504 .
  • the SFU 504 may also be called a switch fabric unit (SFU).
  • SFU switch fabric unit
  • the SFU 504 is used to complete data exchange between the interface boards.
  • the interface board 502 and the interface board 503 may communicate through the switching fabric board 504 .
  • the main control board 501 is coupled to the interface board.
  • the main control board 501, the interface board 502, the interface board 503, and the switching fabric board 504 are connected to the system backplane through the system bus to realize intercommunication.
  • an inter-process communication protocol (inter-process communication, IPC) channel is established between the main control board 501, the interface board 502, and the interface board 503, and the main control board 501, the interface board 502, and the interface board 503 Communicate through IPC channels.
  • IPC inter-process communication
  • the network optimization device includes a control plane and a forwarding plane.
  • the control plane includes the main control board 501 and the central processing unit 5011.
  • the forwarding plane includes various components that perform forwarding, such as the forwarding entry storage 5024, the physical interface card 5023, and the network processing device 5022.
  • the control plane executes routers, generates forwarding tables, processes signaling and protocol packets, configures and maintains the status of network devices, and other functions.
  • the control plane sends the generated forwarding tables to the forwarding plane.
  • the network processor 5022 controls The forwarding table issued above looks up the table and forwards the packets received by the physical interface card 5023.
  • the forwarding table delivered by the control plane may be stored in the forwarding table item storage 5024 .
  • the control plane and the forwarding plane can be completely separated and not on the same network device.
  • main control boards there may be one or more main control boards, and when there are multiple main control boards, it may include the main main control board and the standby main control board. There may be one or more interface boards. The stronger the data processing capability of the network management device, the more interface boards it provides. There may also be one or more physical interface cards on the interface board. There may be no SFU, or there may be one or more SFUs. When there are multiple SFUs, they can jointly implement load sharing and redundant backup. Under the centralized forwarding architecture, the network management device does not need the switching network board, and the interface board is responsible for the processing function of the service data of the whole system.
  • the network management device can have at least one SFU, through which data exchange between multiple interface boards can be realized, and large-capacity data exchange and processing capabilities can be provided. Therefore, the data access and processing capabilities of network management devices with a distributed architecture are greater than those with a centralized architecture.
  • the form of the network management device can also be that there is only one board, that is, there is no switching fabric board, and the functions of the interface board and the main control board are integrated on this board. At this time, the CPU and the main control board on the interface board The central processing unit on the control board can be combined into one central processing unit on the board to perform the superimposed functions of the two.
  • This form of network management equipment has low data exchange and processing capabilities (for example, low-end switches or network devices such as routers). Which architecture to use depends on the specific networking deployment scenario, and there is no limitation here.
  • the embodiment of the present application also provides a computer-readable storage medium, where an instruction is stored in the computer-readable storage medium, and the instruction is executed by a processor to implement the method for determining a network optimization strategy provided by the foregoing method embodiment.
  • the embodiment of the present application also provides a computer program product containing instructions, and when the computer program product is run on a computer, it causes the computer to execute the method for determining the network optimization strategy provided by the above method embodiment.
  • the embodiment of the present application also provides a traffic forwarding system, as shown in FIG. 1 , FIG. 4 and FIG. 10 , the system includes: a controller 01 and a first network device 02 .
  • the structure of the controller 01 may refer to FIG. 11 to FIG. 13 , and the controller 01 is configured to execute the method for determining the network optimization strategy provided by the above method embodiment.
  • the first network device 02 is configured to execute the network optimization policy determined by the controller 01 on the target service flow.
  • the traffic forwarding system may be a wide area network, and the first network device 02 may be an AR.
  • the foregoing embodiments may be fully or partially implemented by software, hardware, firmware or any combination thereof.
  • the above-described embodiments may be implemented in whole or in part in the form of computer program products.
  • the computer program product includes one or more computer instructions.
  • the computer program instructions When the computer program instructions are loaded or executed on the computer, the processes or functions according to the embodiments of the present application will be generated in whole or in part.
  • the computer may be a general purpose computer, a special purpose computer, a computer network, or other programmable devices.
  • the computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions may be transmitted from a website, mobile terminal, computer, server or data
  • the center transmits to another website mobile terminal, computer, server or data center through wired (such as coaxial cable, optical fiber, twisted pair) or wireless (such as infrared, wireless, microwave, etc.).
  • the computer-readable storage medium may be any medium that can be accessed by a computer or a data storage device such as a server or a data center that includes one or more sets of media.
  • the media may be magnetic media (eg, floppy disks, hard disks, magnetic tape), optical media (eg, optical discs), or semiconductor media.
  • the semiconductor medium may be a solid state drive (SSD).

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Physics & Mathematics (AREA)
  • Algebra (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Mathematical Physics (AREA)
  • Probability & Statistics with Applications (AREA)
  • Pure & Applied Mathematics (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)

Abstract

提供了一种网络优化策略的确定方法、装置及系统。网络优化设备能够采用网络优化算法处理网络的性能参数和目标业务流的业务需求，得到目标网络优化策略，并能够对目标业务流应用该目标网络优化策略，以使得目标业务流的传输性能满足该业务需求。由于网络优化设备能够基于网络优化算法自动确定网络优化策略，因此有效提高了网络优化策略的确定效率。又由于在确定网络优化策略时，网络优化设备还考虑了网络的性能参数，因此确定出的网络优化策略能够适应网络性能的动态变化，也即是，在网络性能动态变化时，也能确保目标业务流的传输性能满足业务需求。

Description

网络优化策略的确定方法、装置及系统
本申请要求于2021年9月7日提交的申请号为202111043719.X、发明名称为“自适应广域优化方法和系统”的中国专利申请的优先权，以及要求于2021年10月29日提交的申请号为202111270140.7、发明名称为“网络优化策略的确定方法、装置及系统”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请涉及通信领域，特别涉及一种网络优化策略的确定方法、装置及系统。
背景技术
广域网(wide area network，WAN)是一种连接不同地区的局域网或城域网的通信网络。为了确保业务流在WAN中的传输质量，WAN通常能够提供如下多种网络优化策略：前向纠错(forward error correction，FEC)、压缩、多发选收、逐包负载分担、抖动缓冲、丢包主动重传和传输控制协议(transmission control protocol，TCP)加速等。
相关技术中，运维人员可以基于业务流的业务需求，例如服务等级协议(service level agreement，SLA)中定义的需求，在网络设备中配置能够满足该业务需求的网络优化策略。网络设备进而可以对业务流执行该网络优化策略。例如，若某个业务流的业务需求为高质量和高吞吐率，则运维人员为该业务流配置的网络优化策略可以包括FEC。若某个业务流的业务需求为高效率，则运维人员为该业务流配置的网络优化策略可以包括TCP加速。
但是，上述由运维人员根据业务需求确定并配置网络优化策略的方法的效率较低。
发明内容
本申请提供了一种网络优化策略的确定方法、装置及系统，可以解决网络优化策略配置效率较低的技术问题，技术方案如下：
一方面，提供了一种网络优化策略的确定方法，应用于网络优化设备；该方法包括：获取网络的性能参数，并采用网络优化算法处理输入参数，得到目标网络优化策略，其中，该输入参数包括网络的性能参数和目标业务流的业务需求；之后，对目标业务流应用目标网络优化策略，以使该目标业务流在网络中的传输性能满足业务需求。
由于网络优化设备能基于网络优化算法自动确定网络优化策略，因此有效提高了网络优化策略的确定效率。又由于在确定网络优化策略时，网络优化设备不仅考虑了业务流的业务需求，还考虑了网络的性能参数，因此确定出的网络优化策略能够适应网络性能的动态变化。
可选地，该获取网络的性能参数的过程可以包括：获取目标业务流在网络中将要流经的链路的性能参数。
由于目标业务流将要流经的链路的性能是影响该目标业务流的传输性能的关键因素，因此在确定目标网络优化策略时，可以仅获取网络中该目标业务流将要流经的链路的性能参数。由此，可以在确保确定出的目标网络优化策略的准确性的前提下，减小网络优化设备所需处理的数据的数据量，提高数据处理的效率。
可选地，采用网络优化算法处理网络的性能参数和目标业务流的业务需求，得到目标网络优化策略的过程可以包括：采用网络优化算法处理输入参数，以从多个不同种类的网络优 化技术中选择至少一种网络优化技术，得到目标网络优化策略。
其中，该多个不同种类的网络优化技术可以包括下述技术中的至少两种：FEC技术、压缩技术、多发选收技术、逐包负载分担技术、抖动缓冲技术、丢包主动重传技术和TCP加速技术等。
可选地，该输入参数还可以包括：对目标业务流应用临时网络优化技术后目标业务流的第一临时传输性能，该临时网络优化技术属于该多个不同种类的网络优化技术；相应的，从多个不同种类的网络优化技术中选择至少一种网络优化技术，得到目标网络优化策略的过程可以包括：若第一临时传输性能不满足业务需求，则从该多个不同种类的网络优化技术中，选择与该临时网络优化技术不同的至少一种网络优化技术，得到目标网络优化策略。
网络优化设备对目标业务流应用临时网络优化技术后，还可以基于目标业务流在网络中的实际传输性能，对应用于目标业务流的网络优化技术进行调整，以确保调整后的网络优化技术(即目标网络优化策略)能够使得目标业务流的传输性能满足业务需求。由此，可以实现对应用于目标业务流的网络优化策略的自适应调整，确保在网络性能动态变化的过程中，目标业务流的传输性能始终能够满足业务需求。
可选地，采用网络优化算法处理网络的性能参数和目标业务流的业务需求，得到目标网络优化策略的过程可以包括：采用网络优化算法处理输入参数，以从一种网络优化技术的参数值范围内选择一组参数值，得到目标网络优化策略。
本申请提供的方案中，网络优化设备除了可以选择网络优化技术的种类，还可以选择网络优化技术的参数值，以得到最终应用于目标业务流的目标网络优化策略。
可选地，该输入参数还可以包括：对目标业务流应用参数值为第一临时参数值的一种网络优化技术后，目标业务流的第二临时传输性能；相应的，从一种网络优化技术的参数值范围内选择一组参数值，得到目标网络优化策略的过程可以包括：若第二临时传输性能不满足业务需求，则在一种网络优化技术的参数值范围内，对第一临时参数值进行调整，得到目标网络优化策略。
当目标业务流的临时传输性能不满足业务需求时，网络优化设备还可以对应用于目标业务流的网络优化技术的参数值进行自适应调整。由此，可以确保在网络性能动态变化的过程中，目标业务流的传输性能始终能够满足业务需求。
可选地，输入参数还可以包括：对目标业务流应用参数值为第二临时参数值的一种网络优化技术后，目标业务流的第三临时传输性能；相应的，从一种网络优化技术的参数值范围内选择一组参数值，得到目标网络优化策略的过程可以包括：若第三临时传输性能优于业务需求，则在一种网络优化技术的参数值范围内，对第二临时参数值进行调整，得到目标网络优化策略；其中，执行参数值为第二临时参数值的一种网络优化技术所需的设备资源，多于执行目标网络优化策略所需的设备资源。
当目标业务流的临时传输性能优于业务需求时，网络优化设备也可以对网络优化技术的参数值进行调整，以确保在满足业务需求的前提下，尽量减少网络优化策略所消耗的网络设备的设备资源。由此，可以释放更多的设备资源以对其他业务流执行网络优化策略，即可以有效提高网络设备的资源利用率，增加该网络设备所能够承载的业务流的数量。
可选地，该输入参数还可以包括：用于执行目标网络优化策略的网络设备的资源使用信息；通过在输入参数中增加资源使用信息，可以确保网络设备的设备资源足够执行目标网络优化策略。也即是，可以确保确定出的目标网络优化策略能够被网络设备有效执行。
可选地，网络优化设备采用网络优化算法处理输入参数，得到目标网络优化策略的过程可以包括：将网络性能参数和业务需求输入至优化模型，得到优化模型输出的一种或多个不同种类的网络优化技术；其中，该网络优化算法确定出的目标网络优化策略是从该一种或多个不同种类的网络优化技术中确定的。
例如，网络优化算法还可以包括在线学习算法，网络优化设备可以基于该在线学习算法对优化模型输出的一种或多个不同种类的网络优化技术的种类，和/或，网络优化技术的配置参数的参数值进行选择，得到目标网络优化策略。
可选地，该优化模型可以是对训练样本进行训练得到的，该训练样本可以包括：参考网络优化策略，在对参考业务流应用参考网络优化策略之前网络的性能参数，以及在对参考业务流应用参考网络优化策略后参考业务流的传输性能。
其中，该优化模型可以是由网络优化设备训练得到的，或者可以是由模型训练器训练并下发至网络优化设备的。并且，网络优化设备或模型训练器还可以周期性地对该优化模型进行更新，以改善该优化模型的性能。
可选地，该网络优化设备可以为用于执行目标网络优化策略的网络设备，相应的，对目标业务流应用目标网络优化策略的过程可以包括：对目标业务流执行目标网络优化策略。
可选地，该网络优化设备可以为控制器，相应的，对目标业务流应用目标网络优化策略的过程可以包括：将目标网络优化策略下发至用于执行目标网络优化策略的网络设备，以使得该网络设备对目标业务流执行该目标网络优化策略。
由于每个控制器可以与多个网络设备连接，因此采用控制器作为网络优化设备，可以实现对流经不同网络设备的业务流的网络优化技术的集中计算。并且，便于对该网络优化算法进行维护和更新。
另一方面，提供了一种网络优化设备，该网络优化设备包括至少一个模块，该至少一个模块用于实现上述方面所提供的网络优化策略的确定方法。
又一方面，提供了一种网络优化设备，该网络优化设备包括：存储器，处理器及存储在该存储器上并可在该处理器上运行的计算机程序，该处理器执行该计算机程序时实现如上述方面所提供的网络优化策略的确定方法。
再一方面，提供了一种计算机可读存储介质，该计算机可读存储介质中存储有指令，该指令由处理器执行以实现如上述方面所提供的网络优化策略的确定方法。
再一方面，提供了一种包含指令的计算机程序产品，当该计算机程序产品在计算机上运行时，使得计算机执行上述方面所提供的网络优化策略的确定方法。
再一方面，提供了一种流量转发系统，该系统包括：控制器和网络设备；其中，该控制器用于执行如上述方面提供的网络优化策略的确定方法，该网络设备用于对目标业务流执行该控制器确定出的网络优化策略。
上述方面所提供的网络优化设备、计算机可读存储介质、计算机程序产品以及流量转发系统所获得的技术效果，均与上述方面所提供的网络优化策略的确定方法中对应的技术手段获得的技术效果近似，在这里不再赘述。
综上所述，本申请提供了一种网络优化策略的确定方法、装置及系统。本申请提供的方案中，网络优化设备能够采用网络优化算法处理网络的性能参数和目标业务流的业务需求，得到目标网络优化策略，并能够对目标业务流应用该目标网络优化策略，以使得目标业务流的传输性能满足该业务需求。由于网络优化设备能够基于网络优化算法自动确定网络优化策 略，因此有效提高了网络优化策略的确定效率。又由于在确定网络优化策略时还考虑了网络的性能参数，因此确定出的网络优化策略能够适应网络性能的动态变化。由此，在网络性能动态变化时，也能确保目标业务流的传输性能满足业务需求。
附图说明
图1是本申请实施例提供的一种流量转发系统的结构示意图；
图2是本申请实施例提供的一种网络优化策略的确定方法的流程图；
图3是本申请实施例提供的另一种流量转发系统的结构示意图；
图4是本申请实施例提供的又一种流量转发系统的结构示意图；
图5是本申请实施例提供的一种优化模型的输入参数和输出的网络优化策略的示意图；
图6是本申请实施例提供的一种优化模型的工作原理的示意图；
图7是本申请实施例提供的一种网络优化算法的工作原理的示意图；
图8是本申请实施例提供的一种状态机的示意图；
图9是本申请实施例提供的再一种流量转发系统的结构示意图；
图10是本申请实施例提供的再一种流量转发系统的结构示意图；
图11是本申请实施例提供的一种网络优化设备的结构示意图；
图12是本申请实施例提供的另一种网络优化设备的结构示意图；
图13是本申请实施例提供的再一种网络优化设备的结构示意图。
具体实施方式
下面结合附图详细介绍本申请实施例提供的方案。
图1是本申请实施例提供的一种流量转发系统的结构示意图，如图1所示，该系统(也可以称为网络)可以包括控制器01和多个网络设备。例如，图1中示出的网络设备包括第一网络设备02和第二网络设备03。其中，控制器01与至少一个网络设备(例如第一网络设备02)之间建立有通信连接，各个网络设备之间建立有通信连接。
该控制器01用于对其所连接的各个网络设备进行统一的管理和控制，每个网络设备用于向其他网络设备转发业务流的报文。例如，运维人员可以在控制器01中为目标业务流配置网络优化技术，控制器01可以将该网络优化技术下发至第一网络设备02。第一网络设备02进而可以对目标业务流执行该网络优化技术，并将目标业务流的报文转发至第二网络设备03。
其中，该控制器01可以为一台服务器，或者由若干台服务器组成的服务器集群，或者是一个云计算服务中心。该第一网络设备02和第二网络设备03均可以为路由器或交换机等具有报文转发功能的网络设备，也可以称为转发设备。例如，该第一网络设备02和第二网络设备03均可以为接入路由器(access router，AR)。
可选地，本申请实施例提供的流量转发系统可以是广域网，例如可以是软件定义广域网(software-defined WAN，SD-WAN)。广域网中承载的业务流具有复杂和多变的特点，并且广域网中可配置的网络优化技术(例如FEC、压缩、多发选收、逐包负载分担、抖动缓冲、丢包主动重传和TCP加速等)较多，且每种网络优化技术的配置参数多样。由此，导致运维人员手动配置广域优化技术的效率较低。并且，在网络运行的过程中，可能出现目标业务流所流经的链路的性能变差，导致预先配置的网络优化技术无法满足目标业务流的业务需求的情况。
本申请实施例提供了一种网络优化策略的确定方法，网络优化设备可以采用网络优化算法对网络的性能参数和目标业务流的业务需求进行处理，从而得到能够应用于该目标业务流的目标网络优化策略。其中，该目标网络优化策略可以包括至少一种网络优化技术，或者，可以包括至少一种网络优化技术，以及至少一种网络优化技术的配置参数的参数值。其中，网络优化技术可以是广域优化技术。由于本申请实施例提供的方法无需运维人员手动确定并配置网络优化策略，因此有效提高了网络优化策略确定和配置的效率。又由于在确定网络优化策略时还考虑了网络的性能参数，因此确定出的网络优化策略能够适应网络性能的动态变化。由此，在网络性能动态变化时，也能确保目标业务流的传输性能满足业务需求。
图2是本申请实施例提供的一种网络优化策略的确定方法的流程图，该方法可以应用于网络优化设备。该网络优化设备可以是控制器，也可以是用于转发业务流的网络设备。如图2所示，该方法包括：
步骤101、获取网络的性能参数。
在本申请实施例中，网络的性能参数可以包括下述参数中的至少一种：时延、丢包率、抖动和吞吐量等。并且，网络的性能参数可以是该网络中各个链路的性能参数，或者，可以是目标业务流在该网络中将要流经的链路的性能参数。其中，目标业务流是指需要应用网络优化技术的业务流，目标业务流将要流经的链路的数量可以大于或等于1。若目标业务流将要流经的链路的数量大于1，即目标业务流在网络中具有多条可选链路(也称可行链路或可达链路)，则网络优化设备可以获取每条可选链路的性能参数。
可以理解的是，网络中的每个业务流可以采用五元组(或四元组或七元组)唯一标识。相应的，网络优化设备可以基于目标业务流的唯一标识，确定该目标业务流将要流经的链路。其中，目标业务流将要流经的链路的可以包括专线链路、普通互联网(Internet)链路和移动通信链路等。移动通信链路可以是第四代移动通信技术(4th generation mobile communication technology，4G)链路或5G链路等。
由于目标业务流将要流经的链路的性能是影响该目标业务流的传输性能的关键因素，因此在确定目标网络优化策略时，网络优化设备可以仅获取网络中该目标业务流将要流经的链路的性能参数。由此，可以在确保确定出的目标网络优化策略的准确性的前提下，减小网络优化设备所需处理的数据的数据量，提高数据处理的效率。
可选地，如图3所示，网络中的各个网络设备中均配置有信息采集模块，该信息采集模块可以通过随路测量或主动测量的方式采集网络的性能参数。下文以网络的性能参数为目标业务流将要流经的链路的性能参数为例，并以网络优化设备为第一网络设备02为例进行说明。假设目标业务流将要流经的链路为第一网络设备02和第二网络设备03之间的链路，即第一网络设备02为目标业务流的发送侧的网关设备，第二网络设备03为目标业务流的接收侧的网关设备。则在随路测量时，第一网络设备02中的信息采集模块可以在目标业务流的业务报文中封装标记信息(例如业务报文的发送时间戳和/或业务报文的全局序列号等)，并向第二网络设备03发送该业务报文。在主动测量时，第一网络设备02中的信息采集模块可以主动向第二网络设备03发送专用于测量性能参数的探测报文，该探测报文中封装有标记信息。
第二网络设备02中的信息采集模块接收到封装有标记信息的业务报文或探测报文后，可以基于该标记信息计算链路的性能参数，并将计算得到的性能参数反馈至第一网络设备02。例如，第二网络设备02可以基于报文中的全局序列号计算链路的丢包率和吞吐量，并可以基 于报文中的发送时间戳计算链路的时延。或者，第二网络设备02可以将报文的接收时间戳反馈至第一网络设备02，并由第一网络设备02根据报文的发送时间戳和接收时间戳计算链路的时延。
对于网络优化设备为控制器的场景，第一网络设备01可以向第二网络设备02发送封装有标记信息的业务报文或探测报文。并且，如图4所示，第一网络设备01中的信息采集模块可以向控制器01的信息接收模块发送报文的检测信息，例如报文中封装的标记信息，和/或，封装有标记信息的报文的个数。第二网络设备02中的信息采集模块接收到封装有标记信息的报文后，也可以向控制器01的信息接收模块发送接收到的报文的检测信息，例如报文的接收时间戳，和/或，接收到的封装有标记信息的报文的个数。控制器01的信息接收模块进而可以基于第一网络设备01和第二网络设备02发送的检测信息，计算链路的性能参数。
步骤102、获取网络设备的资源使用信息。
其中，该网络设备至少包括用于对目标业务流执行网络优化技术的第一网络设备，即目标业务流的发送侧的网关设备。可选地，该网络设备还可以包括目标业务流在网络中的接收侧的网关设备(即第二网络设备)。
网络设备的资源使用信息可以包括下述信息中的至少一种：会话(session)数量、处理器使用信息和内存使用信息等。其中，处理器可以包括中央处理器(central processing unit，CPU)、网络处理器(network processing unit，NPU)和图形处理器(graphics processing unit，GPU)中的一种或多种。可以理解的是，该资源使用信息可以采用网络设备已使用的资源量、已使用的资源的百分比，和/或，剩余可用的资源量来表征。
对于网络优化设备为第一网络设备的场景，参考图3，第一网络设备02的信息采集模块可以采集自身的资源使用信息。第二网络设备03的信息采集模块可以将第二网络设备03的资源使用信息发送至第一网络设备02。
对于网络优化设备为控制器的场景，参考图4，第一网络设备02的信息采集模块可以向控制器01的信息接收模块发送该第一网络设备02的资源使用信息。第二网络设备03的信息采集模块可以向控制器01的信息接收模块发送该第二网络设备03的资源使用信息。
步骤103、采用网络优化算法处理输入参数，得到目标网络优化策略，该输入参数包括网络的性能参数、资源使用信息和目标业务流的业务需求。
在本申请实施例中，网络优化设备中预先配置有网络优化算法和需要进行网络优化的目标业务流的业务需求。其中，该业务需求(也可以称为性能需求或服务质量需求)可以是SLA需求或其他约定需求，且该业务需求可以包括对下述至少一种性能参数的需求：时延、丢包率、抖动和吞吐量等。
网络优化设备可以周期性获取网络的性能参数和网络设备的资源使用信息，并可以周期性地采用网络优化算法处理该性能参数、资源使用信息和目标业务流的业务需求，以得到目标网络优化策略。网络优化设备采用网络优化算法处理输入参数的过程中，可以从多个不同种类的网络优化技术中选择至少一种网络优化技术，和/或，从一种网络优化技术的参数值范围内选择一组参数值，从而得到目标网络优化策略。其中，每种网络优化技术具有至少一个配置参数，每种网络优化技术的参数值范围可以包括该种网络优化技术的每个配置参数的可选数值范围。相应的，每种网络优化技术的一组参数值可以包括该种网络优化技术的至少一个配置参数的参数值。基于上述分析可知，该目标网络优化策略可以包括至少一种网络优化技术，或者，该目标网络优化策略可以包括至少一种网络优化技术，以及该至少一种网络优 化技术中，一种或多种网络优化技术的配置参数的参数值。
可选地，该多个不同种类的网络优化技术可以包括FEC技术、压缩技术、多发选收技术、逐包负载分担技术、抖动缓冲技术、丢包主动重传技术、TCP加速技术和七层协议优化技术等。该七层协议优化技术可以包括：超文本传输协议(hyper text transfer protocol，HTTP)优化技术和文件传输协议(file transfer protocol，FTP)优化技术等。其中，FEC技术的配置参数可以包括编码块大小(block size)、丢包率和冗余率等；压缩技术的配置参数可以包括压缩算法的类型和滑动窗口的大小等；多发选收技术的配置参数可以包括重组等待时间和抖动等；逐包负载分担技术的配置参数可以包括各个链路的权重和重组等待时间等；抖动缓冲技术的配置参数可以包括缓存大小和缓存等待时间等；丢包主动重传技术的配置参数可以包括：缓存大小，缓存等待时间，以及发送端报文释放等待时间等；TCP加速技术的配置参数可以包括：拥塞窗口的大小和慢启动阈值等；HTTP优化技术的配置参数可以包括资源缓存策略和数据预取方式等。
其中，编码块大小是指相邻两次冗余包编码操作所需间隔的报文的个数。例如，假设编码块大小为20，则表示每隔20个报文执行一次冗余包编码操作，且每次冗余包编码操作可以生成一个或多个冗余包。HTTP优化技术中的资源缓存策略可以包括最近最少使用(least recently used，LRU)策略和最近最不常用(least frequently used，LFU)策略等。
可以理解的是，若目标业务流在网络中具有多条可选链路，则网络优化设备采用网络优化算法处理输入参数后，不仅可以得到网络优化技术(或网络优化技术及其参数值)，还可以从该多条可选链路中确定出至少一条目标链路。该至少一条目标链路为用于转发该目标业务流的业务报文的链路。
还可以理解的是，网络优化设备也可以不执行上述步骤102，相应的，该输入参数也可以不包括网络设备的资源使用信息。
步骤104、对目标业务流应用目标网络优化策略，以使该目标业务流在网络中的传输性能满足该业务需求。
网络优化设备确定出目标网络优化策略后，即可对目标业务流应用该目标网络优化策略。由于网络优化设备在确定目标网络优化策略时考虑了网络的性能参数，因此即使网络性能动态变化，也可以确保应用该目标网络优化策略后，该目标业务流在网络中的传输性能依旧能够满足该目标业务流的业务需求。若该输入参数还包括网络设备的资源使用信息，则可以确保用于执行该目标网络优化策略的网络设备，具有足够的设备资源以执行该目标网络优化策略。也即是，可以确保确定出的目标网络优化策略能够被网络设备有效执行。
其中，对于网络优化设备为第一网络设备的场景，如图3所示，该第一网络设备02中配置有优化模块和执行模块。该优化模块采用网络优化算法处理输入参数，得到目标网络优化策略后，执行模块可以对目标业务流执行该目标网络优化策略。也即是，该执行模块可以采用目标网络优化策略对目标业务流的业务报文进行处理。
对于网络优化设备为控制器的场景，如图4所示，控制器01中配置有优化模块。该优化模块采用网络优化算法处理输入参数，得到目标网络优化策略后，可以将该目标网络优化策略下发至第一网络设备02。第一网络设备02的执行模块进而可以对目标业务流执行该目标网络优化策略。
由于控制器可以与多个网络设备连接，因此采用控制器作为网络优化设备，可以实现对流经不同网络设备的业务流的网络优化技术的集中计算。并且，便于对该网络优化算法进行 维护和更新。
如图3所示，该第一网络设备02还可以包括流量识别模块、封装模块和转发模块。其中，流量识别模块用于识别需要进行网络优化的目标业务流，例如流量识别模块可以基于业务流的五元组识别该目标业务流。该封装模块用于对目标业务流的业务报文进行封装，例如封装模块可以采用通用路由封装协议(generic routing encapsulation，GRE)在业务报文封装目标网络优化策略的标识。转发模块用于向第二网络设备03转发目标业务流的业务报文，例如，转发模块可以通过至少一条目标链路转发该目标业务流的业务报文。
继续参考图3，第二网络设备03还包括解封装模块、逻辑处理模块和转发模块。其中，解封装模块用于对接收到的业务报文进行解封装。逻辑处理模块用于对业务报文进行逻辑处理，例如可以基于业务报文中封装的目标网络优化策略的标识，采用对应的解码技术对业务报文进行解码。转发模块用于向目标业务流的接收端设备转发逻辑处理后的业务报文。
可选地，在上述步骤104之后，网络优化设备还可以继续执行上述步骤101至步骤103，以更新该目标网络优化策略。例如，网络优化设备可以周期性执行上述步骤101至步骤103；或者，网络优化设备可以在检测到网络的性能参数，和/或，网络设备的资源使用信息发生变化时，再次执行上述步骤101至步骤103。基于此，在网络性能或网络设备的资源使用情况发生变化时，网络优化设备能够自适应调整应用于目标业务流的网络优化策略，以确保能够满足该目标业务流的业务需求。
下文对上述步骤103的实现过程进行介绍。如图5所示，网络优化设备中部署的网络优化算法包括优化模型(也可以称为自适应模型)。网络优化设备可以将获取到的网络的性能参数、网络设备的资源使用信息以及目标业务流的业务需求输入至该优化模型，得到该优化模型输出的一种或多个不同种类的网络优化技术。之后，网络优化设备可以从优化模型输出的一种或多个不同种类的网络优化技术中确定目标网络优化策略。例如，网络优化设备可以随机选择一种网络优化技术，得到目标网络优化策略。
可选地，参考图5，优化模型可以输出至少一个网络优化策略(policy)，每个网络优化策略包括：至少一条目标链路的标识，至少一种网络优化技术，以及每种网络优化技术的配置参数的参数值。例如，如图5所示，该优化模型可以输出n个网络优化策略，该n为大于1的整数。
其中，网络优化策略P1中的目标链路的标识为：链路1，网络优化技术为FEC技术，且该FEC技术的参数值包括：编码块大小为10，丢包率为2％。网络优化策略P2中的网络优化技术包括逐包负载分担技术，目标链路的标识包括：链路1和链路2，即该链路1和链路2用于实现逐包负载分担。该逐包负载分担技术的参数值包括：链路1的权重W1＝2，链路2的权重W2＝1。并且，该网络优化策略P2中的网络优化技术还包括：分别与链路1和链路2对应的FEC技术。其中，与链路1对应的FEC技术的参数值包括：编码块大小为30，丢包率为1％；与链路2对应的FEC技术的参数值包括：编码块大小为20，丢包率为10％。
网络优化策略Pn中的目标链路的标识为：链路K，网络优化技术为多发选收技术，且该多发选收技术的参数值包括：重组等待时间为100毫秒(ms)，抖动为10ms。
可选地，该优化模型可以是基于规则的模型，且对于不同类型的业务流，该优化模型输出的网络优化策略所包含的内容可以不同。示例的，参考图6，对于目标类型的业务流(图6以语音业务的业务流，以及视频等实时业务的业务流为例)，该优化模型可以先基于业务流 的业务需求为该业务流选择目标链路，以及至少一种网络优化技术。之后，优化模型可以基于网络的性能参数(或者性能参数和网络设备的资源使用信息)确定网络优化技术的配置参数的参数值。例如，对于FEC技术，可以根据链路的性能参数确定编码块大小和丢包率。相应的，对于该目标类型的业务流，优化模型输出的网络优化策略可以包括：至少一条目标链路的标识，至少一种网络优化技术，以及每种网络优化技术的配置参数的参数值。
对于除目标类型之外的其他类型的业务流，该优化模型则可以仅基于业务需求为该业务流选择目标链路，而无需再确定网络优化技术。相应的，对于其他类型的业务流，该优化模型输出的网络优化策略可以仅包括至少一条目标链路的标识。
可以理解的是，当网络性能较好时，该优化模型输出的目标类型的业务流的网络优化策略中也可以仅包括至少一条目标链路的标识，而不包括网络优化技术。相应的，网络优化设备无需对目标类型的业务流应用网络优化技术。由此，可以确保在网络性能变好时，网络优化设备能够及时对目标业务流去使能网络优化技术，进而有效释放网络设备的设备资源。
可选地，该优化模型基于输入参数确定网络优化技术的规则，可以是基于不同网络优化技术的特点并结合实验分析确定的。表1中示意性示出了不同网络优化技术的特点，如表1所示，FEC技术是一种通过冗余包来抗丢包和降低重传时延的网络优化技术，该FEC技术在部分场景下还能够提升吞吐。例如，可以提升大带宽时延积(bandwidth-delay product，BDP)丢包链路或非阻塞性丢包率高的链路的吞吐。该FEC技术的使用约束主要为：引入额外带宽抗丢包，不适用阻塞链路。其中，非阻塞性丢包也称为随机性丢包，其是指并非由链路阻塞引起的丢包。
表1
Figure PCTCN2022116441-appb-000001
从表1还可以看出，TCP加速技术、逐包负载分担技术和压缩技术的主要目的均是用于提 升吞吐，多发选收技术的主要目的是实现高可靠。并且，可以理解的是，不同网络优化技术对会话数的限制可以不同。如某个网络优化技术对会话数的限制为M(M为大于1的整数)，则表示网络设备最多能够同时对M个业务流执行该网络优化技术。示例的，FEC技术对会话数的限制M可以为32。
可以理解的是，上述基于规则的优化模型的精度可能难以满足某些复杂网络场景的需求。为了提高确定出的网络优化策略的精度，如图7所示，该网络优化算法还可以包括在线学习算法。该在线学习算法能够基于目标业务流在网络中的传输性能，从优化模型输出的多个不同种类的网络优化技术中选择至少一种网络优化技术，得到目标网络优化策略，和/或，从优化模型输出的一种网络优化技术的参数值范围内选择一组参数值，得到目标网络优化策略。也即是，网络优化设备能够通过在线学习算法，对优化模型输出的网络优化技术的种类，和/或，网络优化技术的参数值进行选择，得到最终的目标网络优化策略。其中，目标业务流在网络中的传输性能可以采用下述至少一种性能参数表征：时延、丢包率、抖动和吞吐量等。
在第一种可能的实现方式中，对于该优化模型输出多个不同种类的网络优化技术的场景，该网络优化算法所处理的输入参数还可以包括：对目标业务流应用临时网络优化技术之后，该目标业务流的第一临时传输性能。其中，该临时网络优化技术属于该多个不同种类的网络优化技术；相应的，网络优化设备通过在线学习算法确定目标网络优化策略的过程可以包括：
若该第一临时传输性能不满足目标业务流的业务需求，则从优化模型输出的多个不同种类的网络优化技术中，选择与该临时网络优化技术不同的至少一种网络优化技术，得到目标网络优化策略。
可选地，网络优化设备可以选择与该临时网络优化技术不同的另一种网络优化技术，得到目标网络优化策略。或者，网络优化设备可以选择与该临时网络优化技术不同的另一种网络优化技术，并将该临时网络优化技术和该另一种网络优化技术组合，得到目标网络优化策略。也即是，网络优化设备可以对目标业务流应用多种网络优化技术。
示例的，假设对于某个目标业务流，优化模型输出的网络优化技术包括FEC技术和多发选收技术。若网络优化设备将FEC技术作为临时网络优化技术应用于该目标业务流之后，该目标业务流在网络中的第一临时传输性能不满足业务需求，则网络优化设备可以将多发选收技术确定为目标网络优化策略，并对该目标业务流应用该多发选收技术。或者，网络优化设备可以将FEC技术和多发选收技术的组合确定为目标网络优化策略，并对该目标业务流分别应用FEC技术和多发选收技。
在该第一种可能的实现方式中，由于网络优化设备对目标业务流应用临时网络优化技术后，还可以基于目标业务流在网络中的实际传输性能(即临时传输性能)，对应用于目标业务流的网络优化技术进行调整，因此可以确保调整后的网络优化技术(即目标网络优化策略)能够使得目标业务流的传输性能满足业务需求。由此，可以实现对应用于目标业务流的网络优化策略的自适应调整，确保在网络性能动态变化的过程中，目标业务流的传输性能始终能够满足业务需求。
在第二种可能的实现方式中，对于该优化模型输出一种网络优化技术的场景，该网络优化算法所处理的输入参数还可以包括：对目标业务流应用参数值为第一临时参数值的该种网络优化技术后，该目标业务流的第二临时传输性能。相应的，网络优化设备通过在线学习算法确定目标网络优化策略的过程可以包括：
若该第二临时传输性能不满足目标业务流的业务需求，则在该种网络优化技术的参数值 范围内，对该第一临时参数值进行调整，得到目标网络优化策略。
其中，网络优化技术的参数值范围可以是优化模型输出的，且该参数值范围可以包括网络优化技术的每个配置参数的可选数值范围。网络优化设备在确定第二临时传输性能不满足目标业务流的业务需求后，可以在至少一个配置参数的参数值的可选数值范围内，调节该配置参数的参数值，直至目标业务流的传输性能满足业务需求。
例如，假设优化模型输出的一种网络优化技术为FEC技术，该FEC技术的配置参数包括编码块大小和丢包率，则优化模型输出的FEC技术的参数值范围可以包括：编码块大小的可选数值范围，以及丢包率的可选数值范围。相应的，网络优化设备若确定第二临时传输性能不满足目标业务流的业务需求，则可以调节编码块大小和/或丢包率的大小，直至目标业务流的传输性能满足业务需求。
作为一种可选的示例，网络优化设备在调节配置参数的参数值时，可以先分别按照增大参数值和减小参数值的方向对配置参数的参数值进行调节，并分别获取对目标业务流应用增大参数值的网络优化技术后，该目标业务流的第一更新传输性能，以及对目标业务流应用减小参数值的网络优化技术后，该目标业务流的第二更新传输性能。之后，网络优化设备可以采用效益函数(例如马氏距离)分别判断第一更新传输性能与该目标业务流的业务需求的第一距离，以及第二更新传输性能与该目标业务流的业务需求的第二距离。
由于距离越近则表示更新传输性能越逼近业务需求，因此若第一距离小于第二距离，则网络优化设备可以在配置参数的可选数值范围内，继续增大参数值，直至对目标业务流应用增大参数值的网络优化技术后，该目标业务流的传输性能满足业务需求。若第二距离小于第一距离，则网络优化设备可以在配置参数的可选数值范围内，继续减小参数值，直至对目标业务流应用减小参数值的网络优化技术后，该目标业务流的传输性能满足业务需求。
作为另一种可选的示例，网络优化设备在调节配置参数的参数值时，可以先按照目标方向对配置参数的参数值进行调节，并获取对目标业务流应用调节参数值的网络优化技术后，该目标业务流的更新传输性能。若该更新传输性能相比于第二临时传输性能更逼近目标业务流的业务需求，则网络优化设备可以继续按照该目标方向调节参数值，直至对目标业务流应用调节参数值的网络优化技术后，该目标业务流的传输性能满足业务需求。若该更新传输性能相比于第二临时传输性能远离目标业务流的业务需求，则网络优化设备可以按照与该目标方向相反的方向调节参数值，直至对目标业务流应用调节参数值的网络优化技术后，该目标业务流的传输性能满足业务需求。其中，该目标方向可以为增大参数值的方向或者减小参数值的方向。
在上述两种示例中，网络优化设备均可以按照固定的调整比例对配置参数的参数值进行调整。例如，网络优化设备每次可以将参数值增大为原参数值的1.25倍，或者，减小为原参数值的0.75倍。
可以理解的是，若网络优化设备在将一种网络优化技术的配置参数的参数值调节至可选数值范围的上限或下限后，仍无法满足目标业务流的业务需求，则网络优化设备可以重新采用网络优化算法处理输入参数，以获取目标网络优化策略。
还可以理解的是，对于优化模型输出多种网络优化技术的场景，网络优化设备可以结合上述第一种可能的实现方式和第二种可能的实现方式来确定目标网络优化策略。例如，网络优化设备若确定目标业务流的第一临时传输性能不满足业务需求，则可以先采用上述第二种可能的实现方式对该临时网络优化技术的配置参数的参数值进行调整。若将该临时网络优化 技术的配置参数的参数值调节至可选数值范围的上限或下限后，仍无法满足业务需求，则网络优化设备可以采用上述第一种可能的实现方式，选择与该临时网络优化技术不同的至少一种网络优化技术，得到目标网络优化策略。
示例的，如图7所示，假设优化模型输出的网络优化技术包括P1和P2两种，则网络优化设备可以先将网络优化技术P1作为临时网络优化技术，并应用于目标业务流。之后，网络优化设备可以检测该目标业务流在网络中的第一临时传输性能U1。若该第一临时传输性能U1不满足目标业务流的业务需求T，则网络优化设备可以将网络优化技术P1的配置参数的参数值由A 1调节为A 2，并应用于目标业务流。然后，网络优化设备可以检测该目标业务流在网络中的第二临时传输性能U2。若该第二临时传输性能U2仍不满足业务需求T，则网络优化设备可以继续对网络优化技术P1的配置参数的参数值进行调整。
若网络优化设备将网络优化技术P1的配置参数的参数值调整至上限A k后，该目标业务流的临时传输性能仍不满足业务需求T，则网络优化设备可以对目标业务流应用网络优化技术P2。若应用网络优化技术P2后，目标业务流的传输性能满足该业务需求T，则网络优化设备可以将该网络优化技术P2作为目标网络优化策略并继续应用于该目标业务流。其中，图7中的i为正整数，k为大于1的整数，Ui表示网络优化设备第i次获取到的目标业务流的传输性能。
可选地，在该第二种可能的实现方式中，该输入参数还可以包括：对目标业务流应用参数值为第二临时参数值的一种网络优化技术后，该目标业务流的第三临时传输性能；网络优化设备通过在线学习算法确定目标网络优化策略的过程可以包括：
若该第三临时传输性能优于目标业务流的业务需求，则在该种网络优化技术的参数值范围内，对该第二临时参数值进行调整，得到目标网络优化策略。其中，执行参数值为该第二临时参数值的该种网络优化技术所需的设备资源，多于执行该目标网络优化策略所需的设备资源。
也即是，对于第三临时传输性能优于业务需求的场景，网络优化设备也可以对网络优化技术的第二临时参数值进行调整，以确保在满足业务需求的前提下，尽量减少网络优化策略所消耗的网络设备的设备资源。由此，可以释放更多的设备资源以对其他业务流执行网络优化策略，即可以有效提高网络设备的资源利用率，增加该网络设备所能够承载的业务流的数量(即所支持的session数量)。
可以理解的是，网络优化设备可以按照上文所述的两种可选的示例，对第二临时参数值进行调整，以使得对目标业务流应用调整参数值后的网络优化技术之后，目标业务流的传输性能满足该业务需求，且更逼近该业务需求。
在本申请实施例中，网络优化设备中还配置有状态机，网络优化设备在通过在线学习算法确定目标网络优化策略的过程中，可以根据该状态机调整网络优化技术的参数值或切换网络优化技术的种类。下文以图8所示的状态机为例，对网络优化设备确定目标网络优化策略的过程进行介绍。
如图8所示，在初始化(start up)状态下，网络优化设备能够基于优化模型的输出，确定初始的网络优化技术(即临时网络优化技术)。在步骤S1中，网络优化设备可以对目标业务流应用该临时网络优化技术。之后，网络优化设备进入持续优化(continue optimization)状态，并每隔调整间隔执行一次步骤S2，其中该调整间隔的时长可以为1秒。在步骤S2中，网络优化设备可以对临时网络优化技术的配置参数的参数值进行调整，并检测调整参数值之后，该目标业务流的临时传输性能是否满足业务需求。
在步骤S3中，网络优化设备若检测到前一个状态(pre-state)为初始化状态，且当前目标业务流的临时传输性能不满足业务需求，则可以执行步骤S4。在步骤S4中，网络优化设备可以选择与临时网络优化技术不同的至少一种网络优化技术。
在步骤S5中，网络优化设备若检测到前一个状态为持续优化状态，且当前目标业务流的临时传输性能不满足业务需求，则可以进入回滚状态，并执行步骤S6。在该步骤S6中，网络优化设备可以将临时网络优化技术的配置参数的参数值恢复至上一次调整之前的数值。也即是，网络优化设备在持续调整参数值的过程中，若前若干次调整后，目标业务流的临时传输性能均能满足业务需求，但最后一次调整后，目标业务流的临时传输性能不再满足业务需求。则网络优化设备可以确定：最后一次调整之前，目标业务流的临时传输性能已逼近业务需求，因此可以将临时网络优化技术的配置参数的参数值回滚至最后一次调整之前的数值，从而得到目标网络优化策略。
如图8所示，在步骤S6之后，网络优化设备可以进入稳定(stable)状态，并执行步骤S7。在步骤S7中，网络优化设备可以每隔检查间隔，执行一次持续优化的操作，即每隔检查间隔调整一次参数值，并检测调整参数值之后，目标业务流的传输性能是否满足业务需求。其中，该检查间隔的时长大于调整间隔的时长，例如该检查间隔可以为5秒或10秒。若调整参数值之后，目标业务流的传输性能不满足业务需求(即持续优化失败)，则网络优化设备可以执行步骤S8。在步骤S8中，网络优化设备可以将当前选定的网络优化技术的配置参数的参数值回滚至最后一次调整之前的数值，并可以增大该检查间隔。例如，可以将该检查间隔增大为原来的两倍。之后，网络优化设备即可按照增大后的检测间隔执行持续优化的操作。
继续参考图8，网络优化设备在稳定状态下，若通过步骤S9检测到目标业务流所流经的目标链路故障(down)，或者目标链路的性能持续恶化，则可以进入重置(reset)状态，并执行步骤S11。在步骤S11中，网络优化设备可以重新选择与临时网络优化技术不同的至少一种网络优化技术。从图8可以看出，网络优化设备在持续优化状态下，若通过步骤S10检测到目标业务流所流经的目标链路故障或者性能持续恶化，则也可以进入重置状态。
可选地，在本申请实施例中，该优化模型可以是对多个训练样本进行训练得到的。例如可以是采用机器学习算法(如决策树或贝叶斯网络等)或强化学习算法训练得到的。其中，每个训练样本可以包括：参考网络优化策略，在对参考业务流应用该参考网络优化策略之前该网络的性能参数，以及在对该参考业务流应用该参考网络优化策略后该参考业务流的传输性能。
其中，该参考网络优化策略可以包括至少一种网络优化技术，或者可以包括至少一种网络优化技术，以及该至少一种网络优化技术中一种或多种网络优化技术的参数值。其中，该至少一种网络优化技术可以包括FEC技术、压缩技术、多发选收技术、逐包负载分担技术、抖动缓冲技术、丢包主动重传技术和/或TCP加速技术等。不同训练样本中的参考业务流可以是不同类型的应用的业务流，例如可以是语音业务的业务流或视频业务的业务流。
示例的，如图9和图10所示，流量转发系统中还可以包括模型训练器04，该模型训练器04可以为一台服务器，或者由若干台服务器组成的服务器集群，或者是一个云计算服务中心。该模型训练器04可以用于对多个训练样本进行训练得到的优化模型，并将优化模型下发至网络优化设备。例如，参考图9，对于网络优化设备为第一网络设备02的场景，模型训练器04可以将优化模型下发至第一网络设备02中的优化模块。参考图10，对于网络优化设备为控制器01的场景，模型训练器04可以将优化模型下发至控制器01中的优化模块。
参考图9和图10可以看出，用于训练模型的各个训练样本可以是各个网络设备中的信息采集模块采集并上报至模型训练器04的，或者，可以是控制器01收集并上报至模型训练器04的。
可以理解的是，该模型训练器04在训练得到优化模型后，还可以周期性的获取训练样本，并基于获取到的训练样本对该优化模型进行更新。之后，模型训练器04可以将更新后的优化模型下发至网络优化设备，网络优化设备进而可以采用更新后的优化模型确定网络优化策略。
还可以理解的是，该模型训练器04也可以集成在网络优化设备中，即网络优化设备也具有模型训练和模型更新的功能。
还可以理解的是，上述网络优化算法也可以是采用机器学习算法或深度学习算法训练得到的网络优化模型。也即是，网络优化设备将输入参数输入至网络优化模型后，即可得到该网络优化模型输出的目标网络优化策略。
可选地，本申请实施例提供的方法的步骤的先后顺序可以进行适当调整，步骤也可以根据情况进行相应增减。例如，步骤102可以在步骤101之前执行，或者网络优化设备也可以无需执行步骤102。
综上所述，本申请实施例提供了一种网络优化策略的确定方法。网络优化设备能够采用网络优化算法处理网络的性能参数和目标业务流的业务需求，得到目标网络优化策略，并能够对目标业务流应用该目标网络优化策略，以使得目标业务流的传输性能满足该业务需求。由于网络优化设备能够基于网络优化算法自动确定网络优化策略，因此有效提高了网络优化策略的确定效率。又由于在确定网络优化策略时还考虑了网络的性能参数，因此确定出的网络优化策略能够适应网络性能的动态变化。由此，在网络性能动态变化时，也能确保目标业务流的传输性能满足业务需求。
其中，在网络性能变差导致目标业务流的传输性能无法满足业务需求时，网络优化设备能够通过调整网络优化技术的参数值，切换网络优化技术的种类，和/或，叠加其他网络优化技术来改善该目标业务流的传输性能，以确保满足该目标业务流的业务需求。在网络性能变好时，网络优化设备能够通过调整网络优化技术的参数值，或者，去使能网络优化技术来释放网络设备的资源，以提高网络设备的资源利用率。
本申请实施例还提供了一种网络优化设备，该网络优化设备可以应用于图1、图3、图4、图9或图10所示的系统中，且可以用于实现上述方法实施例提供的方法。如图11所示，该网络优化设备30包括：
获取模块301，用于获取网络的性能参数。该获取模块301的功能实现可以参考上述方法实施例中步骤101的相关描述。
优化模块302，用于采用网络优化算法处理输入参数，得到目标网络优化策略，输入参数包括网络的性能参数和目标业务流的业务需求。该优化模块302的功能实现可以参考上述方法实施例中步骤103的相关描述。
应用模块303，用于对目标业务流应用目标网络优化策略，以使目标业务流在网络中的传输性能满足业务需求。该应用模块303的功能实现可以参考上述方法实施例中步骤104的相关描述。
可选地，该获取模块301，用于获取目标业务流在网络中将要流经的链路的性能参数。
可选地，该优化模块302，可以用于从多个不同种类的网络优化技术中选择至少一种网络优化技术，得到目标网络优化策略。
可选地，该输入参数还包括：对目标业务流应用临时网络优化技术后目标业务流的第一临时传输性能，临时网络优化技术属于多个不同种类的网络优化技术；该优化模块302，可以用于若第一临时传输性能不满足业务需求，则从多个不同种类的网络优化技术中，选择与临时网络优化技术不同的至少一种网络优化技术，得到目标网络优化策略。
可选地，该优化模块302，可以用于采用网络优化算法处理输入参数，以从一种网络优化技术的参数值范围内选择一组参数值，得到目标网络优化策略。
可选地，该输入参数还包括：对目标业务流应用参数值为第一临时参数值的一种网络优化技术后，目标业务流的第二临时传输性能；该优化模块302，可以用于若第二临时传输性能不满足业务需求，则在一种网络优化技术的参数值范围内，对第一临时参数值进行调整，得到目标网络优化策略。
可选地，该输入参数还包括：对目标业务流应用参数值为第二临时参数值的一种网络优化技术后，目标业务流的第三临时传输性能；该优化模块302，可以用于若第三临时传输性能优于业务需求，则在一种网络优化技术的参数值范围内，对第二临时参数值进行调整，得到目标网络优化策略；
其中，执行参数值为第二临时参数值的一种网络优化技术所需的设备资源，多于执行目标网络优化策略所需的设备资源。
可选地，该输入参数还包括：用于执行目标网络优化策略的网络设备的资源使用信息；该资源使用信息用于使网络设备的设备资源足够执行目标网络优化策略。该获取模块301的功能实现还可以参考上述方法实施例中步骤102的相关描述。
可选地，该优化模块302，可以用于将网络性能参数和业务需求输入至优化模型，得到优化模型输出的一种或多个不同种类的网络优化技术；
其中，目标网络优化策略从该一种或多个不同种类的网络优化技术中确定。
可选地，该优化模型是对训练样本进行训练得到的，训练样本包括：参考网络优化策略，在对参考业务流应用参考网络优化策略之前网络的性能参数，以及在对参考业务流应用参考网络优化策略后参考业务流的传输性能。
作为一种可能的实现方式，该网络优化设备30可以为控制器，该应用模块303可以用于将目标网络优化策略下发至用于执行该目标网络优化策略的网络设备。该获取模块301可以是上述实施例中的信息接收模块。
作为另一种可能的实现方式，该网络优化设备30可以为用于执行该目标网络优化策略的网络设备，该应用模块303可以用于对目标业务流执行目标网络优化策略。该获取模块301可以是上述实施例中的信息采集模块。
综上所述，本申请实施例提供了一种网络优化设备，该网络优化设备能够采用网络优化算法处理网络的性能参数和目标业务流的业务需求，得到目标网络优化策略，并能够对目标业务流应用该目标网络优化策略，以使得目标业务流的传输性能满足该业务需求。由于网络优化设备能够基于网络优化算法自动确定网络优化策略，因此有效提高了网络优化策略的确定效率。又由于在确定网络优化策略时还考虑了网络的性能参数，因此确定出的网络优化策略能够适应网络性能的动态变化。由此，在网络性能动态变化时，也能确保目标业务流的传输性能满足业务需求。
所属领域的技术人员可以清楚地了解到，为描述的方便和简洁，上文描述的网络优化设备及各模块的具体工作过程，可以参考前述方法实施例中的对应过程，在此不再赘述。
应理解的是，本申请实施例提供的网络优化设备可以用专用集成电路(application-specific integrated circuit，ASIC)实现，或可编程逻辑器件(programmable logic device，PLD)实现，上述PLD可以是复杂程序逻辑器件(complex programmable logical device，CPLD)，现场可编程门阵列(field-programmable gate array，FPGA)，通用阵列逻辑(generic array logic，GAL)或其任意组合。或者，也可以通过软件实现上述方法实施例提供的网络优化策略的确定方法，当通过软件实现上述方法实施例提供的网络优化策略的确定方法时，上述网络优化设备中的各个模块也可以为软件模块。
图12是本申请实施例提供的另一种网络优化设备的结构示意图，该网络优化设备可以应用于图1、图3、图4、图9或图10所示的系统中，且可以用于实现上述方法实施例提供的方法。参考图12，该网络优化设备可以包括：处理器401、存储器402、网络接口403和总线404。其中，总线404用于连接处理器401、存储器402和网络接口403。通过网络接口403(可以是有线或者无线)可以实现与其他设备之间的通信连接。存储器402中存储有计算机程序4021，该计算机程序4021用于实现各种应用功能。
应理解，在本申请实施例中，处理器401可以是CPU，该处理器401还可以是其他通用处理器、数字信号处理器(digital signal processing，DSP)、专用集成电路(application-specific integrated circuit，ASIC)、现场可编程门阵列(field-programmable gate array，FPGA)、通用阵列逻辑(generic array logic，GAL)或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件等，或上述处理器的任意组合。通用处理器可以是微处理器或者是任何常规的处理器等。
存储器402可包括易失性存储器，非易失性存储器，或其组合。其中，非易失性存储器可以包括只读存储器(read-only memory，ROM)、可编程只读存储器(programmable ROM，PROM)、可擦除可编程只读存储器(erasable PROM，EPROM)、电可擦除可编程只读存储器(electrically EPROM，EEPROM)、闪存或其任意组合。易失性存储器可以包括随机存取存储器(random access memory，RAM)，例如静态随机存取存储器(static RAM，SRAM)、动态随机存取存储器(DRAM)、同步动态随机存取存储器(synchronous DRAM，SDRAM)、双倍数据速率同步动态随机存取存储器(double data date SDRAM，DDR SDRAM)、增强型同步动态随机存取存储器(enhanced SDRAM，ESDRAM)、同步连接动态随机存取存储器(synchlink DRAM，SLDRAM)、直接内存总线随机存取存储器(direct rambus RAM，DR RAM)或其任意组合。
总线404除包括数据总线之外，还可以包括电源总线、控制总线和状态信号总线等。但是为了清楚说明起见，在图中将各种总线都标为总线404。
处理器401可以被配置为执行存储器402中存储的计算机程序4021来实现上述方法实施例提供的网络优化策略的确定方法。
图13是本申请实施例提供的又一种网络优化设备的结构示意图，该网络优化设备可以应用于图1、图3、图4、图9或图10所示的系统中，且可以用于实现上述方法实施例提供的方法。如图13所示，该网络优化设备包括主控板501和接口板502。
主控板501也称为主处理单元(main processing unit，MPU)或路由处理卡(route processor card)，主控板501用于对网络优化设备中各个组件进行控制和管理，包括路由计算、设备管 理、设备维护和协议处理等功能。主控板501包括：中央处理器5011和存储器5012。
接口板502也称为线路接口单元卡(line processing unit，LPU)、线卡(line card)或业务板。接口板502用于提供各种业务接口并实现业务帧的转发。业务接口包括而不限于以太网接口、基于SONET/SDH的业务帧(packet over SONET/SDH，POS)接口等。SONET是指同步光纤网络(synchronous optical network)，SDH是指同步数字体系(synchronous digital hierarchy)。以太网接口例如是灵活以太网业务接口(flexible Ethernet clients，FlexE Clients)。接口板502包括：中央处理器5021、网络处理器5022、转发表项存储器5024和物理接口卡(physical interface card，PIC)5023。
接口板502上的中央处理器5021用于对接口板502进行控制管理并与主控板501上的中央处理器5011进行通信。
网络处理器5022用于实现报文的转发处理。网络处理器5022的形态可以是转发芯片。转发芯片可以是网络处理器(network processor，NP)。在一些实施例中，转发芯片可以通过ASIC或FPGA实现。具体而言，网络处理器5022用于基于转发表项存储器5024保存的转发表转发接收到的报文，如果报文的目的地址为网络优化设备的地址，则将该报文上送至CPU(如中央处理器5021)处理；如果报文的目的地址不是网络优化设备的地址，则根据该目的地址从转发表中查找到该目的地址对应的下一跳和出接口，将该报文转发到该目的地址对应的出接口。其中，上行报文的处理可以包括：报文入接口的处理，转发表查找；下行报文的处理可以包括：转发表查找等等。在一些实施例中，中央处理器也可执行转发芯片的功能，比如基于通用CPU实现软件转发，从而接口板中不需要转发芯片。
物理接口卡5023用于实现物理层的对接功能，原始的流量由此进入接口板502，以及处理后的报文从该物理接口卡5023发出。物理接口卡5023也称为子卡，可安装在接口板502上，负责将光电信号转换为报文并对报文进行合法性检查后转发给网络处理器5022处理。在一些实施例中，中央处理器5021也可执行网络处理器5022的功能，比如基于通用CPU实现软件转发，从而物理接口卡5023中不需要网络处理器5022。
可选地，网络优化设备包括多个接口板，例如网络优化设备还包括接口板503，接口板503包括：中央处理器5031、网络处理器5032、转发表项存储器5034和物理接口卡5033。接口板503中各部件的功能和实现方式与接口板502相同或相似，在此不再赘述。
可选地，网络优化设备还包括交换网板504。交换网板504也可以称为交换网板单元(switch fabric unit，SFU)。在网络管理设备有多个接口板的情况下，交换网板504用于完成各接口板之间的数据交换。例如，接口板502和接口板503之间可以通过交换网板504通信。
主控板501和接口板耦合。例如。主控板501、接口板502和接口板503，以及交换网板504之间通过系统总线与系统背板相连实现互通。在一种可能的实现方式中，主控板501和接口板502及接口板503之间建立进程间通信协议(inter-process communication，IPC)通道，主控板501和接口板502及接口板503之间通过IPC通道进行通信。
在逻辑上，网络优化设备包括控制面和转发面，控制面包括主控板501和中央处理器5011，转发面包括执行转发的各个组件，比如转发表项存储器5024、物理接口卡5023和网络处理器5022。控制面执行路由器、生成转发表、处理信令和协议报文、配置与维护网络设备的状态等功能，控制面将生成的转发表下发给转发面，在转发面，网络处理器5022基于控制面下发的转发表对物理接口卡5023收到的报文查表转发。控制面下发的转发表可以保存在转发表项存储器5024中。在有些实施例中，控制面和转发面可以完全分离，不在同一网络设备上。
值得说明的是，主控板可能有一块或多块，有多块的时候可以包括主用主控板和备用主控板。接口板可能有一块或多块，网络管理设备的数据处理能力越强，提供的接口板越多。接口板上的物理接口卡也可以有一块或多块。交换网板可能没有，也可能有一块或多块，有多块的时候可以共同实现负荷分担冗余备份。在集中式转发架构下，网络管理设备可以不需要交换网板，接口板承担整个系统的业务数据的处理功能。在分布式转发架构下，网络管理设备可以有至少一块交换网板，通过交换网板实现多块接口板之间的数据交换，提供大容量的数据交换和处理能力。所以，分布式架构的网络管理设备的数据接入和处理能力要大于集中式架构的网络管理设备。可选地，网络管理设备的形态也可以是只有一块板卡，即没有交换网板，接口板和主控板的功能集成在该一块板卡上，此时接口板上的中央处理器和主控板上的中央处理器在该一块板卡上可以合并为一个中央处理器，执行两者叠加后的功能，这种形态网络管理设备的数据交换和处理能力较低(例如，低端交换机或路由器等网络设备)。具体采用哪种架构，取决于具体的组网部署场景，此处不做任何限定。
本申请实施例还提供了一种计算机可读存储介质，该计算机可读存储介质中存储有指令，该指令由处理器执行以实现上述方法实施例提供的网络优化策略的确定方法。
本申请实施例还提供了一种包含指令的计算机程序产品，当该计算机程序产品在计算机上运行时，使得计算机执行上述方法实施例提供的网络优化策略的确定方法。
本申请实施例还提供了一种流量转发系统，如图1、图4和图10所示，该系统包括：控制器01和第一网络设备02。其中，控制器01的结构可以参考图11至图13，且该控制器01用于执行上述方法实施例提供的网络优化策略的确定方法。该第一网络设备02用于对目标业务流执行该控制器01确定出的网络优化策略。示例的，该流量转发系统可以为广域网，该第一网络设备02可以为AR。
上述实施例，可以全部或部分地通过软件、硬件、固件或其任意组合来实现。当使用软件或固件实现时，上述实施例可以全部或部分地以计算机程序产品的形式实现。所述计算机程序产品包括一个或多个计算机指令。在计算机上加载或执行所述计算机程序指令时，全部或部分地产生按照本申请实施例所述的流程或功能。所述计算机可以为通用计算机、专用计算机、计算机网络、或者其他可编程装置。所述计算机指令可以存储在计算机可读存储介质中，或者从一个计算机可读存储介质向另一个计算机可读存储介质传输，例如，所述计算机指令可以从一个网站移动终端、计算机、服务器或数据中心通过有线(例如同轴电缆、光纤、双绞线)或无线(例如红外、无线、微波等)方式向另一个网站移动终端、计算机、服务器或数据中心进行传输。所述计算机可读存储介质可以是计算机能够存取的任何介质或者是包含一个或多个介质集合的服务器、数据中心等数据存储设备。所述介质可以是磁性介质(例如，软盘、硬盘、磁带)、光介质(例如，光盘)、或者半导体介质。半导体介质可以是固态硬盘(solid state drive，SSD)。
本申请中术语“至少一个”的含义是指一个或多个，本申请中术语“多个”的含义是指两个或两个以上，本文中术语“系统”和“网络”经常可互换使用。在本文中提及的“和/或”，表示可以存在三种关系，例如，A和/或B，可以表示：单独存在A，同时存在A和B，单独存在B这三种情况。字符“/”一般表示前后关联对象是一种“或”的关系。
以上所述仅为本申请的可选实施例，并不用以限制本申请。本申请的保护范围以权利要求为准。

Claims (27)

  1. 一种网络优化策略的确定方法，其特征在于，应用于网络优化设备；所述方法包括：
    获取网络的性能参数；
    采用网络优化算法处理输入参数，得到目标网络优化策略，所述输入参数包括所述网络的性能参数和目标业务流的业务需求；
    对所述目标业务流应用所述目标网络优化策略，以使所述目标业务流在所述网络中的传输性能满足所述业务需求。
  2. 根据权利要求1所述的方法，其特征在于，所述获取网络的性能参数包括：
    获取所述目标业务流在所述网络中将要流经的链路的性能参数。
  3. 根据权利要求1或2所述的方法，其特征在于，所述采用网络优化算法处理所述网络的性能参数和目标业务流的业务需求，得到目标网络优化策略包括：
    采用网络优化算法处理输入参数，以从多个不同种类的网络优化技术中选择至少一种网络优化技术，得到目标网络优化策略。
  4. 根据权利要求3所述的方法，其特征在于，所述输入参数还包括：对所述目标业务流应用临时网络优化技术后所述目标业务流的第一临时传输性能，所述临时网络优化技术属于所述多个不同种类的网络优化技术；
    所述从多个不同种类的网络优化技术中选择至少一种网络优化技术，得到目标网络优化策略，包括：
    若所述第一临时传输性能不满足所述业务需求，则从所述多个不同种类的网络优化技术中，选择与所述临时网络优化技术不同的至少一种网络优化技术，得到目标网络优化策略。
  5. 根据权利要求1至4任一所述的方法，其特征在于，所述采用网络优化算法处理所述网络的性能参数和目标业务流的业务需求，得到目标网络优化策略包括：
    采用网络优化算法处理输入参数，以从一种网络优化技术的参数值范围内选择一组参数值，得到目标网络优化策略。
  6. 根据权利要求5所述的方法，其特征在于，所述输入参数还包括：对所述目标业务流应用参数值为第一临时参数值的所述一种网络优化技术后，所述目标业务流的第二临时传输性能；
    所述从一种网络优化技术的参数值范围内选择一组参数值，得到目标网络优化策略包括：
    若所述第二临时传输性能不满足所述业务需求，则在所述一种网络优化技术的参数值范围内，对所述第一临时参数值进行调整，得到目标网络优化策略。
  7. 根据权利要求5所述的方法，其特征在于，所述输入参数还包括：对所述目标业务流应用参数值为第二临时参数值的所述一种网络优化技术后，所述目标业务流的第三临时传输性能；
    所述从一种网络优化技术的参数值范围内选择一组参数值，得到目标网络优化策略包括：
    若所述第三临时传输性能优于所述业务需求，则在所述一种网络优化技术的参数值范围内，对所述第二临时参数值进行调整，得到目标网络优化策略；
    其中，执行参数值为所述第二临时参数值的所述一种网络优化技术所需的设备资源，多于执行所述目标网络优化策略所需的设备资源。
  8. 根据权利要求1至7任一所述的方法，其特征在于，所述输入参数还包括：用于执行所述目标网络优化策略的网络设备的资源使用信息；
    所述资源使用信息用于使所述网络设备的设备资源足够执行所述目标网络优化策略。
  9. 根据权利要求1至8任一所述的方法，其特征在于，所述采用网络优化算法处理输入参数，得到目标网络优化策略，包括：
    将所述网络性能参数和所述业务需求输入至优化模型，得到所述优化模型输出的一种或多个不同种类的网络优化技术；
    其中，所述目标网络优化策略从所述一种或多个不同种类的网络优化技术中确定。
  10. 根据权利要求9所述的方法，其特征在于，所述优化模型是对训练样本进行训练得到的，所述训练样本包括：参考网络优化策略，在对参考业务流应用所述参考网络优化策略之前所述网络的性能参数，以及在对所述参考业务流应用所述参考网络优化策略后所述参考业务流的传输性能。
  11. 根据权利要求1至10任一所述的方法，其特征在于，所述网络优化设备为控制器，所述对所述目标业务流应用所述目标网络优化策略，包括：
    将所述目标网络优化策略下发至用于执行所述目标网络优化策略的网络设备。
  12. 根据权利要求1至10任一所述的方法，其特征在于，所述网络优化设备为用于执行所述目标网络优化策略的网络设备，所述对所述目标业务流应用所述目标网络优化策略，包括：
    对所述目标业务流执行所述目标网络优化策略。
  13. 一种网络优化设备，其特征在于，所述网络优化设备包括：
    获取模块，用于获取网络的性能参数；
    优化模块，用于采用网络优化算法处理输入参数，得到目标网络优化策略，所述输入参数包括所述网络的性能参数和目标业务流的业务需求；
    应用模块，用于对所述目标业务流应用所述目标网络优化策略，以使所述目标业务流在所述网络中的传输性能满足所述业务需求。
  14. 根据权利要求13所述的设备，其特征在于，所述获取模块，用于获取所述目标业务流在所述网络中将要流经的链路的性能参数。
  15. 根据权利要求13或14所述的设备，其特征在于，所述优化模块，用于：
    采用网络优化算法处理输入参数，以从多个不同种类的网络优化技术中选择至少一种网 络优化技术，得到目标网络优化策略。
  16. 根据权利要求15所述的设备，其特征在于，所述输入参数还包括：对所述目标业务流应用临时网络优化技术后所述目标业务流的第一临时传输性能，所述临时网络优化技术属于所述多个不同种类的网络优化技术；所述优化模块，用于：
    若所述第一临时传输性能不满足所述业务需求，则从所述多个不同种类的网络优化技术中，选择与所述临时网络优化技术不同的至少一种网络优化技术，得到目标网络优化策略。
  17. 根据权利要求13至16任一所述的设备，其特征在于，所述优化模块，用于：
    采用网络优化算法处理输入参数，以从一种网络优化技术的参数值范围内选择一组参数值，得到目标网络优化策略。
  18. 根据权利要求17所述的设备，其特征在于，所述输入参数还包括：对所述目标业务流应用参数值为第一临时参数值的所述一种网络优化技术后，所述目标业务流的第二临时传输性能；所述优化模块，用于：
    若所述第二临时传输性能不满足所述业务需求，则在所述一种网络优化技术的参数值范围内，对所述第一临时参数值进行调整，得到目标网络优化策略。
  19. 根据权利要求17所述的设备，其特征在于，所述输入参数还包括：对所述目标业务流应用参数值为第二临时参数值的所述一种网络优化技术后，所述目标业务流的第三临时传输性能；所述优化模块，用于：
    若所述第三临时传输性能优于所述业务需求，则在所述一种网络优化技术的参数值范围内，对所述第二临时参数值进行调整，得到目标网络优化策略；
    其中，执行参数值为所述第二临时参数值的所述一种网络优化技术所需的设备资源，多于执行所述目标网络优化策略所需的设备资源。
  20. 根据权利要求13至19任一所述的设备，其特征在于，所述输入参数还包括：用于执行所述目标网络优化策略的网络设备的资源使用信息；
    所述资源使用信息用于使所述网络设备的设备资源足够执行所述目标网络优化策略。
  21. 根据权利要求13至20任一所述的设备，其特征在于，所述优化模块，用于：
    将所述网络性能参数和所述业务需求输入至优化模型，得到所述优化模型输出的一种或多个不同种类的网络优化技术；
    其中，所述目标网络优化策略从所述一种或多个不同种类的网络优化技术中确定。
  22. 根据权利要求21所述的设备，其特征在于，所述优化模型是对训练样本进行训练得到的，所述训练样本包括：参考网络优化策略，在对参考业务流应用所述参考网络优化策略之前所述网络的性能参数，以及在对所述参考业务流应用所述参考网络优化策略后所述参考业务流的传输性能。
  23. 根据权利要求13至22任一所述的设备，其特征在于，所述网络优化设备为控制器，所述应用模块，用于：
    将所述目标网络优化策略下发至用于执行所述目标网络优化策略的网络设备。
  24. 根据权利要求13至22任一所述的设备，其特征在于，所述网络优化设备为用于执行所述目标网络优化策略的网络设备，所述应用模块，用于：
    对所述目标业务流执行所述目标网络优化策略。
  25. 一种网络优化设备，其特征在于，所述设备包括：存储器，处理器及存储在所述存储器上并可在所述处理器上运行的计算机程序，所述处理器执行所述计算机程序时实现如权利要求1至12任一所述的方法。
  26. 一种计算机可读存储介质，其特征在于，所述计算机可读存储介质中存储有指令，所述指令由处理器执行以实现如权利要求1至12任一所述的方法。
  27. 一种流量转发系统，其特征在于，所述系统包括：控制器和网络设备；
    其中，所述控制器用于执行如权利要求1至11任一所述的网络优化策略的确定方法，所述网络设备用于对目标业务流执行所述控制器确定出的网络优化策略。
PCT/CN2022/116441 2021-09-07 2022-09-01 网络优化策略的确定方法、装置及系统 WO2023036042A1 (zh)

Applications Claiming Priority (4)

Application Number Priority Date Filing Date Title
CN202111043719.X 2021-09-07
CN202111043719 2021-09-07
CN202111270140.7A CN115776443A (zh) 2021-09-07 2021-10-29 网络优化策略的确定方法、装置及系统
CN202111270140.7 2021-10-29

Publications (1)

Publication Number Publication Date
WO2023036042A1 true WO2023036042A1 (zh) 2023-03-16

Family

ID=85388397

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/116441 WO2023036042A1 (zh) 2021-09-07 2022-09-01 网络优化策略的确定方法、装置及系统

Country Status (2)

Country Link
CN (1) CN115776443A (zh)
WO (1) WO2023036042A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116996397A (zh) * 2023-09-27 2023-11-03 之江实验室 一种网络丢包优化的方法、装置、存储介质及电子设备

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116996895B (zh) * 2023-09-27 2024-01-02 香港中文大学（深圳） 一种基于深度强化学习的全网时延和吞吐率联合优化方法

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101087464A (zh) * 2007-07-04 2007-12-12 华为技术有限公司 网络优化方法和装置
CN103763205A (zh) * 2014-01-14 2014-04-30 合肥工业大学 硅通孔负载全局均衡的三维片上网络延迟上界优化方法
CN111294819A (zh) * 2018-12-07 2020-06-16 中国移动通信集团陕西有限公司 一种网络优化方法及装置
WO2021007019A1 (en) * 2019-07-08 2021-01-14 Google Llc Optimizing a cellular network using machine learning
CN113015147A (zh) * 2021-02-20 2021-06-22 Oppo广东移动通信有限公司 一种网络优化方法及终端设备
CN113329430A (zh) * 2021-06-15 2021-08-31 中国联合网络通信集团有限公司 网络优化方法和装置

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101087464A (zh) * 2007-07-04 2007-12-12 华为技术有限公司 网络优化方法和装置
CN103763205A (zh) * 2014-01-14 2014-04-30 合肥工业大学 硅通孔负载全局均衡的三维片上网络延迟上界优化方法
CN111294819A (zh) * 2018-12-07 2020-06-16 中国移动通信集团陕西有限公司 一种网络优化方法及装置
WO2021007019A1 (en) * 2019-07-08 2021-01-14 Google Llc Optimizing a cellular network using machine learning
CN113015147A (zh) * 2021-02-20 2021-06-22 Oppo广东移动通信有限公司 一种网络优化方法及终端设备
CN113329430A (zh) * 2021-06-15 2021-08-31 中国联合网络通信集团有限公司 网络优化方法和装置

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116996397A (zh) * 2023-09-27 2023-11-03 之江实验室 一种网络丢包优化的方法、装置、存储介质及电子设备
CN116996397B (zh) * 2023-09-27 2024-01-09 之江实验室 一种网络丢包优化的方法、装置、存储介质及电子设备

Also Published As

Publication number Publication date
CN115776443A (zh) 2023-03-10

Similar Documents

Publication Publication Date Title
WO2023036042A1 (zh) 网络优化策略的确定方法、装置及系统
EP3278514B1 (en) Data transmission
WO2018001062A1 (zh) 一种计算转发路径的方法及网络设备
US7835285B2 (en) Quality of service, policy enhanced hierarchical disruption tolerant networking system and method
US10135736B1 (en) Dynamic trunk distribution on egress
Muthumanikandan et al. Link failure recovery using shortest path fast rerouting technique in SDN
US9425893B1 (en) Methods and apparatus for implementing optical integrated routing with traffic protection
US11234159B2 (en) Systems and methods for congestion control on mobile edge networks
CN116233256A (zh) 调度路径配置方法、装置、设备及存储介质
CN115460114A (zh) 业务功能链拥塞反馈
CN113438182B (zh) 一种基于信用的流量控制系统和流量控制方法
US20240089204A1 (en) Communication Method, Device, and System
Ndao et al. Optimal placement of virtualized DUs in O-RAN architecture
CN111431812B (zh) 一种报文转发控制方法及装置
CN114095448A (zh) 一种拥塞流的处理方法及设备
Utsumi et al. A new explicit congestion notification scheme for satellite IP networks
Qin et al. Edge computing aided congestion control using neuro-dynamic programming in NDN
Farahmand et al. A closed-loop rate-based contention control for optical burst switched networks
WO2023093804A1 (zh) 一种丢包管理方法及相关装置
US20240056885A1 (en) Multi-access traffic management
EP4277226A1 (en) Packet transmission method, transmission control method, apparatus, and system
WO2023040783A1 (zh) 获取能力、发送能力信息的方法、装置、系统及存储介质
Wang et al. Congestion detection and link control via feedback in RDMA transmission
Boanea et al. Performance evaluation of a situation aware multipath routing solution
Kazemi et al. An IoT-based packet aggregation mechanism for the SDN-based wide area networks

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22866506

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 2022866506

Country of ref document: EP

ENP Entry into the national phase

Ref document number: 2022866506

Country of ref document: EP

Effective date: 20240313

NENP Non-entry into the national phase

Ref country code: DE