Network control method and system

Abstract

The invention discloses a network control method and system, and belongs to the technical field of network management of a wide area network, and the method comprises the steps: obtaining key network data of the wide area network through a public cloud node; determining a network optimization decision for the wide area network according to the key network data based on a target agent; the network optimization decision is used for improving the network communication performance of the wide area network; determining network configuration information of network equipment in the wide area network according to the network optimization decision; and sending the network configuration information to corresponding network equipment in the wide area network to control the corresponding network equipment to execute network optimization processing corresponding to the network configuration information.

Description

Technical Field

[0001] The present application belongs to the technical field of network management of wide area networks, and in particular, relates to a network control method and system. Background Art

[0002] With the rapid expansion of global business and the increasing complexity of enterprise IT (Information Technology) architecture, Software-Defined Networking in a Wide Area Network (SD-WAN) technology has emerged as a key technology to support enterprise network agility, cost-effectiveness and application performance. SD-WAN technology allows enterprises to manage network connections and policies in a software-defined manner, which not only improves network flexibility but also reduces the cost of relying on traditional hardware equipment.

[0003] Although SD-WAN technology has provided significant improvements in network traffic management, as the amount of data increases and the real-time requirements of applications increase, the existing technology faces challenges in handling high latency and ensuring data transmission security. Especially in multinational enterprise operations, data must be transmitted over long distances, and conventional SD-WAN technology has failed to effectively address the impact of network latency on business performance. Summary of the invention

[0004] To this end, this application discloses the following technical solutions:

[0005] A network control method, comprising:

[0006] Obtain key network data of the WAN through public cloud nodes;

[0007] Based on the target intelligent agent, determining a network optimization decision for the wide area network according to the key network data; the network optimization decision is used to improve the network communication performance of the wide area network;

[0008] Determining network configuration information for network devices in the wide area network according to the network optimization decision;

[0009] The network configuration information is sent to corresponding network devices in the wide area network to control the corresponding network devices to perform network optimization processing corresponding to the network configuration information.

[0010] Optionally, the obtaining of key network data of the wide area network includes:

[0011] Obtaining at least one of network traffic data, performance indicator data, and security log data of the wide area network collected by various data collection units deployed in the wide area network;

[0012] Wherein, each data collection unit is deployed on a point-of-entry device and / or a switch of the wide area network; the key network data includes at least one of the network traffic data, the performance indicator data and the security log data.

[0013] Optionally, the point of presence device includes a cloud access point device deployed at least on a public cloud of the wide area network, and the cloud access point device on the public cloud is used to form a network access layer of the public cloud;

[0014] The obtaining of at least one of the network traffic data, performance indicator data, and security log data of the wide area network collected by each data collection unit deployed in the wide area network comprises at least one of the following:

[0015] Obtaining network traffic data of the wide area network obtained by a data acquisition unit deployed on the cloud access point device by monitoring the traffic flowing through the cloud access point device;

[0016] Obtaining performance indicator data of the wide area network obtained by a data acquisition unit deployed on the cloud access point device by monitoring the link performance of the link to which the cloud access point device is connected;

[0017] A data collection unit deployed on the cloud access point device is obtained, and the security log data of the wide area network is obtained by linkage with the corresponding security device in the wide area network; the security device is a device responsible for network security.

[0018] Optionally, the target agent-based method of determining a network optimization decision for the wide area network according to the key network data includes:

[0019] inputting at least one of the network traffic data, the performance indicator data, and the security log data into an artificial intelligence orchestrator deployed in a central data processing unit of the wide area network;

[0020] Obtain a routing adjustment strategy and / or a network resource reconfiguration strategy for the wide area network generated and output by the artificial intelligence orchestrator based on input data.

[0021] Optionally, determining the network configuration information of the network device in the wide area network according to the network optimization decision includes at least one of the following:

[0022] Based on a network controller disposed in the central data processing unit, generating a forwarding table matching the routing adjustment strategy;

[0023] Based on the network controller, a network resource configuration command matching the network resource reconfiguration strategy is generated; the network configuration information of the network device in the wide area network includes the forwarding table and / or the network resource configuration command;

[0024] The sending of the network configuration information to a corresponding network device in the wide area network to control the corresponding network device to perform network optimization processing corresponding to the network configuration information includes:

[0025] The forwarding table and / or the network resource configuration command are sent to a corresponding point-of-presence device or switch in the wide area network through the network controller, so as to control the corresponding point-of-presence device or switch to perform path selection and data packet forwarding based on the received forwarding table, and / or to perform network resource configuration based on the received network resource configuration command;

[0026] The network resource configuration command includes at least one of the following: a bandwidth allocation configuration command, a quality of service policy configuration command, a load balancing configuration command, a failover configuration command, and a security policy configuration command.

[0027] Optionally, the network control method further includes:

[0028] Obtain execution result data and / or network optimization effect data generated based on the network optimization processing, and adjust the parameters of the target intelligent agent based on the execution result data and / or the network optimization effect data.

[0029] A network control system, comprising:

[0030] Data collection unit, used to collect key network data of the wide area network;

[0031] A central data processing unit is used to: determine a network optimization decision for the wide area network based on the target intelligent agent and the key network data, wherein the network optimization decision is used to improve the network communication performance of the wide area network; determine network configuration information for network devices in the wide area network based on the network optimization decision; and send the network configuration information to corresponding network devices in the wide area network to control the corresponding network devices to perform network optimization processing corresponding to the network configuration information.

[0032] Optionally, the data collection unit is deployed on a point-of-entry device and / or a switch of the wide area network, and is used to collect at least one of network traffic data, performance indicator data and security log data of the wide area network;

[0033] The key network data includes at least one of the network traffic data, the performance indicator data and the security log data.

[0034] Optionally, the central data processing unit includes an artificial intelligence orchestrator and a network controller;

[0035] The artificial intelligence orchestrator is used to generate and output a routing adjustment strategy and / or a network resource reconfiguration strategy for the wide area network based on at least one of the network traffic data, the performance indicator data and the security log data;

[0036] The network controller is configured to: generate a forwarding table matching the route adjustment strategy, and / or generate a network resource configuration command matching the network resource reconfiguration strategy; and send the forwarding table and / or the network resource configuration command to a corresponding point-of-presence device or switch in the wide area network, so as to control the corresponding point-of-presence device or switch to perform path selection and data packet forwarding based on the received forwarding table, and / or perform network resource configuration based on the received network resource configuration command;

[0037] Among them, the network configuration information of the network device in the wide area network includes the forwarding table and / or the network resource configuration command; the network resource configuration command includes at least one of the following: bandwidth allocation configuration command, quality of service policy configuration command, load balancing configuration command, failover configuration command, security policy configuration command.

[0038] Optionally, the central data processing unit is also used to: obtain execution result data and / or network optimization effect data generated based on the network optimization processing, and adjust the parameters of the target intelligent agent based on the execution result data and / or network optimization effect data. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related technologies, the drawings required for use in the embodiments or the related technical descriptions are briefly introduced below. Obviously, the drawings described below are merely embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0040] Figure 1 It is a flow chart of the network control method provided by this application;

[0041] Figure 2 This is a schematic diagram of the traditional SD-WAN architecture;

[0042] Figure 3 This is a schematic diagram of the AI-driven SD-WAN network architecture provided by this application;

[0043] Figure 4 It is another flow chart of the network control method provided by the present application;

[0044] Figure 5 It is an implementation process diagram of the artificial intelligence-driven SD-WAN system provided by this application;

[0045] Figure 6 It is a composition structure diagram of the network control system provided by this application. DETAILED DESCRIPTION

[0046] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0047] The embodiments of the present application provide a network control method and system for implementing an AI-driven network management and control solution by integrating AI technology into the network management of a wide area network, thereby significantly improving the data transmission efficiency and security of the network.

[0048] The network control method provided in the embodiment of the present application is applicable to but not limited to intelligent and efficient network management and optimization of software-defined wide area networks, namely SD-WANs, to ensure the real-time, stability and security of wide area networks such as SD-WANs.

[0049] See also Figure 1 As shown in the method flow chart, the network control method provided in the embodiment of the present application may include the following steps 101 to 104, and these steps are described in detail below.

[0050] Step 101: Obtain key network data of the wide area network through the public cloud node.

[0051] The wide area network may be, but is not limited to, a SD-WAN.

[0052] Specifically, at least one of the network traffic data, performance indicator data and security log data of the wide area network collected by various data collection units deployed in the wide area network can be obtained as the key network data.

[0053] That is, the key network data includes at least one of the network traffic data, performance index data and security log data of the wide area network, but is not limited thereto. For example, it may also include but is not limited to the device status data in the wide area network.

[0054] The performance indicator data may include, but are not limited to, real-time performance parameters such as WAN bandwidth usage, delay, packet loss rate, and security threat indicators.

[0055] Each data collection unit is deployed on a point-of-presence device and / or a switch of a wide area network. The point-of-presence device includes a cloud access point (PoP) device deployed at least on a public cloud of the wide area network. The cloud access point device is also referred to as a PoP node in the embodiment of the present application.

[0056] The traditional technology does not include cloud access point devices in the wide area network architecture. In the embodiment of the present application, cloud access point devices are pre-deployed at least on the public cloud of the wide area network. In addition, cloud access point devices can also be deployed on private clouds, such as deploying PoP nodes on public clouds and private clouds around the world in SD-WAN to form a network access layer with wide coverage. A data collection unit is set on each PoP node to collect key network data in real time.

[0057] In addition, optionally, the embodiment of the present application also deploys data collection units on switches of the wide area network public cloud and private cloud (such as SD-WAN switches) to combine the data collection units deployed on PoP nodes and switches to ensure that key network data such as network traffic data, performance indicator data, and security event logs of the wide area network can be captured in real time and comprehensively.

[0058] Among them, PoP nodes have basic data collection and packet forwarding capabilities, which are used to collect key network data and other required data in real time and forward data packets. These PoP nodes can not only serve as data collection points in the network, but also act as data preprocessing centers to perform some data preprocessing operations, such as cleaning and standardizing the collected data, removing redundant or unnecessary data, etc., to enhance the efficiency and security of data collection.

[0059] The data collection process of switches such as SD-WAN switches and PoP nodes is similar. Optionally, switches focus on collecting local forwarding data of network traffic, while PoP nodes can be mainly responsible for collecting cross-regional link data.

[0060] SD-WAN and other wide area networks also deploy network controllers such as SD-WAN controllers and data processing centers. The network controller is used to centrally manage and configure all PoP nodes and switches. Specifically, the corresponding controller software, such as SD-WAN controller software, can be installed and configured on the network controller to ensure that the network controller can receive data collected from the data acquisition unit in real time to upload data to the data processing center and perform necessary network configuration updates. PoP nodes are uniformly managed and coordinated through the network controller to ensure the effective use and efficient management of network resources.

[0061] The data processing center can be a data center provided by an enterprise or cloud service. When establishing a data processing center in a wide area network, high-performance servers and sufficient storage facilities can be deployed, and advanced artificial intelligence technologies can be integrated, such as installing AI (Artificial Intelligence) data analysis software to receive and process key network data collected from PoP nodes and / or switches throughout the wide area network.

[0062] See also Figure 2 and Figure 3 ,in, Figure 2 It is the architecture of traditional SD-WAN, which is an On-Premises-Overlay network architecture. On-Premises-Overlay network architecture refers to a layer of virtualized network built on the physical network within the enterprise data center. This architecture usually relies on centralized data processing for network decision-making and management. This structure has problems such as low efficiency, high latency, and low security when processing globally distributed data.

[0063] Figure 3 It is a solution provided in the embodiment of the present application. By adding PoP nodes on the wide area network and integrating artificial intelligence technology, an artificial intelligence-based SD-WAN network architecture is formed. This architecture is an On-PoP-Overlay network architecture. By setting a data acquisition unit on each PoP node, key network data such as network traffic, performance indicators and security event logs can be captured in real time, and intelligent data processing and decision-making capabilities can be provided through the artificial intelligence technology integrated in the data processing center. The collaborative work between the data processing center and the PoP nodes based on the integrated artificial intelligence technology can realize efficient data transmission and processing, improve the network response speed and reduce latency. At the same time, based on the efficient AI model, the data is deeply analyzed to identify network traffic patterns, predict potential network congestion and security threats, and improve network security.

[0064] Figure 2 and Figure 3 Headquarter means headquarters, premises means business place, ISP means Internet Service Provider, CPE means Customer Premises Equipment, Private Cloud means private cloud, and Data Centre means data center.

[0065] During implementation, PoP nodes can be widely distributed on the public cloud and private cloud of the wide area network, such as deploying PoP nodes around the world in SD-WAN, to collect and upload large-scale critical network data in real time to ensure the timeliness and accuracy of data collection and transmission.

[0066] Data collection devices such as PoP nodes and switches can transmit the collected critical network data to the data processing center in real time through secure communication protocols for intelligent processing and decision-making.

[0067] Step 102: Based on the target intelligent agent, determine a network optimization decision for the wide area network according to the key network data; the network optimization decision is used to improve the network communication performance of the wide area network.

[0068] The data processing center may be, but is not limited to, a cloud data center.

[0069] The data processing center is deployed with a central data processing unit, and the central data processing unit is further deployed with an artificial intelligence orchestrator and the network controller, such as an SD-WAN path selection AI orchestrator and an SD-WAN controller.

[0070] The target intelligent agent is an artificial intelligence orchestrator deployed on the central data processing unit, such as an SD-WAN path selection AI orchestrator.

[0071] The central data processing unit needs to have strong computing and storage capabilities to process the large amount of critical network data transmitted from PoP nodes and switches. The artificial intelligence orchestrator is responsible for data analysis and AI model training, and the network controller configures and manages the network based on the decisions of the artificial intelligence orchestrator.

[0072] In this step, when determining the network optimization decision for the wide area network based on the target intelligent agent and the key network data, the key network data such as network traffic data, performance indicator data and security log data can be first input into the artificial intelligence orchestrator, and then the artificial intelligence orchestrator analyzes the network status in real time by processing and analyzing the input data, and identifies the patterns and potential problems in the network, and then generates the corresponding network optimization decision. The generated network optimization decision is used to improve the network communication performance of the wide area network, including but not limited to improving the network communication performance of the wide area network in terms of data transmission efficiency, network reliability, security, etc.

[0073] An artificial intelligence orchestrator, or AI orchestrator, can generate and output network optimization decisions for the wide area network by processing and analyzing key network data such as input network traffic data, performance indicator data, and security log data. The generated network optimization decisions include but are not limited to routing adjustment strategies and / or network resource reconfiguration strategies. For example, an artificial intelligence orchestrator determines the optimal path selection strategy by analyzing factors such as network status, bandwidth utilization, and latency, and then generates a routing adjustment strategy that matches the optimal path selection strategy.

[0074] The network optimization decisions such as routing adjustment strategies and / or network resource reconfiguration strategies generated and output by the artificial intelligence orchestrator are a set of high-level strategies that cannot be directly understood and executed by the device. Taking routing adjustment strategies as an example, they may include but are not limited to which POP node to which POP node has the optimal path, which links should be used, etc.

[0075] The routing adjustment strategy is used to adjust the transmission path of data in the network to achieve path optimization, improve transmission rate and reduce latency. The network resource reconfiguration strategy includes but is not limited to the artificial intelligence orchestrator analyzing key network data such as network traffic data, performance indicator data and security log data to identify patterns and potential problems in the network, and then determining bandwidth allocation strategies, QoS (quality of service) strategies, load balancing, failover and / or security strategies based on this.

[0076] For example, the SD-WAN path selection AI orchestrator, based on insights gained through deep learning analysis, uses its decision generation unit to automatically formulate and issue network optimization decisions that match the input key network data, including but not limited to route optimization, bandwidth allocation, failover strategy, and security measures adjustment and optimization, etc. All decisions are aimed at improving network performance, ensuring business continuity or network security.

[0077] The decision-making unit in the AI ​​orchestrator mainly performs route optimization based on the optimized path information analyzed by the AI ​​orchestrator, but it is not limited to this. In addition to route optimization, the decision-making unit can also make decisions based on other real-time network status information, such as combining network bandwidth utilization, latency, packet loss rate, security events and other information to make failover decisions, security protection decisions, etc. Information such as network bandwidth utilization, latency, packet loss rate, and security events is obtained from PoP nodes and SD-WAN switches through the data acquisition unit to further improve the accuracy of optimization decisions.

[0078] In an SD-WAN network, a large number of service data packets may be involved, and the starting and destination nodes of these data packets are often different. To cope with this complexity, the AI ​​orchestrator can predict the optimal network path between each pair of starting and destination nodes. However, in actual applications, not all paths must be predicted by the AI ​​orchestrator. According to the actual situation, the AI ​​orchestrator mainly focuses on the key paths that need to be optimized.

[0079] Specifically, the AI ​​orchestrator will calculate the global optimal path for certain key paths of primary concern based on multiple parameters such as the link status between all POP nodes, WAN bandwidth usage, latency, packet loss rate, etc. These paths may be targeted at business needs with large data traffic.

[0080] For non-critical paths, the AI ​​orchestrator can give the local optimal path based on the link status of each POP node to provide some optional path sets and allow the SD-WAN controller to select the appropriate path to forward business data packets.

[0081] Decisions on network resource reconfiguration strategies such as failover and security protection also rely on the AI ​​orchestrator’s intelligent analysis of key network data such as real-time network performance and event log information reported by POP nodes or switches.

[0082] Taking failover and security protection strategies as an example, the specific decision-making process is as follows:

[0083] 1) Failover strategy: The AI ​​model of the AI ​​orchestrator continuously monitors the health status of each node and link in the network. When it detects that the link performance has degraded or a PoP node has failed, the model will generate new path decisions based on previously learned experience data and automatically transfer network traffic to the best performing or backup link. This strategy is mainly based on real-time information such as packet loss rate, latency, and bandwidth usage, and the AI ​​model automatically selects the appropriate fault recovery path.

[0084] 2) Security protection strategy: For security protection strategy, the AI ​​model analyzes security event logs and threat detection information, such as suspicious traffic patterns or known attack patterns. Based on this information, the decision generation unit in the AI ​​model generates security protection measures, including modifying access control lists (ACLs), adjusting firewall rules, enabling or disabling specific ports, and other measures. The specific strategy is determined by the AI ​​model's analysis of historical security event data, automatically generated in combination with the current network status, and deployed by the SD-WAN controller.

[0085] When multiple POP nodes or switches report the collected key network data to the data processing center, it is optional and not directly input into the AI ​​orchestrator. After the data is reported, the following processing can be performed on it:

[0086] 1) Data preprocessing: First, the key network data reported by the POP nodes will be cleaned and standardized to remove redundant or unnecessary data to ensure that the AI ​​model processes accurate and valid information.

[0087] 2) Data aggregation and analysis: The key network data reported by different POP nodes will be aggregated so that the AI ​​orchestrator can analyze the network status from a global perspective. For example, the data reported by different POP nodes on the same link can be compared to further extract key information (such as link latency, bandwidth utilization, etc.) to provide more effective input for subsequent AI model calculations.

[0088] 3) Feature extraction: The system extracts features closely related to network optimization such as path selection / optimization from the preprocessed data. Only these feature data will eventually be input into the AI ​​orchestrator for optimization calculations.

[0089] The AI ​​model used by the AI ​​orchestrator can be a supervised machine learning model or a reinforcement learning model based on deep learning. There is no restriction on this. The specific model can be determined based on the system's needs and complexity for path optimization.

[0090] Step 103: Determine network configuration information for network devices in the wide area network according to the network optimization decision.

[0091] After obtaining the network optimization decision, a network controller such as an SD-WAN controller can be further used to convert the network optimization decision into network configuration information that can be understood and executed by devices in the network.

[0092] Specifically, the network controller may be used to generate a forwarding table that matches the routing adjustment policy, and / or to generate a network resource configuration command that matches the network resource reconfiguration policy.

[0093] The network resource configuration command may include, but is not limited to, one or more of a bandwidth allocation configuration command, a quality of service policy configuration command, a load balancing configuration command, a failover configuration command, and a security policy configuration command.

[0094] It is easy to understand that the network configuration information of the network device in the wide area network includes the forwarding table and / or the network resource configuration command.

[0095] That is to say, the AI ​​orchestrator generates and outputs high-level network optimization decisions such as routing adjustment strategies and / or network resource reconfiguration strategies by analyzing key network data reported by POP nodes in real time. Network controllers such as SD-WAN controllers are responsible for concretizing the high-level strategies generated by the AI ​​orchestrator and converting them into configuration commands that network devices can understand and execute, such as forwarding tables, QoS policies, load balancing rules, etc., and sending them to POP nodes and related SD-WAN switch devices.

[0096] The network controller will decompose the routing adjustment policy into the specific routing information and data forwarding rules that each POP node / switch needs to execute, thereby forming the forwarding table to achieve the conversion of the high-level routing adjustment policy to an executable forwarding table. In addition to the conversion of the routing adjustment policy, the SD-WAN controller is also responsible for the conversion of the network resource reconfiguration policy, which is mainly reflected in the following aspects:

[0097] 1) Bandwidth allocation: The SD-WAN controller will dynamically convert the bandwidth allocation policy output by the AI ​​orchestrator into the corresponding bandwidth allocation configuration command. For example, for certain high-priority business traffic, the controller can convert the corresponding bandwidth allocation policy into a configuration command to instruct to allocate more bandwidth to a specific link (the link corresponding to the high-priority business traffic) to ensure the rational use of network resources.

[0098] 2) QoS (Quality of Service) Policy: The SD-WAN controller can apply different QoS policies to different types of traffic and convert them into understandable QoS commands. For example, higher priority and better link conditions are assigned to delay-sensitive applications (such as video conferencing and financial transactions), while suboptimal paths are selected for less sensitive traffic.

[0099] 3) Load balancing: The SD-WAN controller converts the load balancing decision output by the AI ​​model based on the path analysis results into corresponding load balancing configuration commands to load balance different POP nodes / switches, ensuring that the traffic in the network is not concentrated on a certain link to avoid network bottlenecks.

[0100] 4) Failover: When certain links or POP nodes fail, the SD-WAN controller can quickly redistribute traffic to ensure business continuity by converting the failover strategy preset by the AI ​​orchestrator into corresponding commands. This involves enabling backup paths, reallocating resources, and quickly switching POP nodes.

[0101] 5) Security policy: The SD-WAN controller can convert the security policies for security threats or abnormal traffic analyzed by the AI ​​orchestrator into corresponding commands to dynamically adjust firewall rules, intrusion detection / prevention system configuration, etc. to improve network security.

[0102] In general, the AI ​​orchestrator is responsible for generating high-level network optimization decisions, while network controllers such as the SD-WAN controller are responsible for implementing these decisions into specific network resource configuration and routing rules to achieve traffic forwarding and dynamic adjustment of resources.

[0103] Step 104: Send the network configuration information to corresponding network devices in the wide area network to control the corresponding network devices to perform network optimization processing corresponding to the network configuration information.

[0104] After using the network controller to generate the forwarding table and the network resource configuration command, the network controller may continue to send the forwarding table and / or the network resource configuration command to the corresponding point of entry device or switch in the wide area network, for example, to the POP node or switch, so as to control the POP node or switch to perform path selection and data packet forwarding based on the received forwarding table, and / or perform network resource configuration based on the received network resource configuration command.

[0105] Exemplarily, after converting the network optimization decision into corresponding forwarding tables and various network resource configuration commands such as bandwidth allocation configuration commands, quality of service policy configuration commands, load balancing configuration commands, failover configuration commands and / or security policy configuration commands, the SD-WAN controller sends the forwarding tables and / or various network resource configuration commands to the PoP nodes and SD-WAN switches to configure the PoP nodes and SD-WAN switches to achieve optimized paths, ensure that network traffic is transmitted along the optimal path, and execute related bandwidth allocation, quality of service policy control, load balancing, failover, security control, etc., to achieve network optimization in terms of bandwidth allocation, quality of service, load balancing, failover, security control, etc., and ensure network stability and security.

[0106] In summary, the network control method provided in the embodiment of the present application collects key network data of the wide area network by utilizing POP nodes and / or switches and other devices, and integrates artificial intelligence technology in the network management of the wide area network, and utilizes a target intelligent agent based on artificial intelligence technology to determine network optimization decisions based on key network data, and converts the network optimization decisions into network configuration information that can be understood and executed by network devices, and then performs network optimization processing based on the network configuration information. It can realize real-time intelligent path optimization that dynamically adapts to changes in the network environment, improve the transmission efficiency and real-time performance of network data, and meet the application's requirements for network real-time performance. At the same time, it can identify patterns and potential problems in the network through analysis of key network data to ensure the stability and security of the network.

[0107] In an alternative embodiment, see Figure 4 The method flow chart shown in the figure, the network control method provided by the present application may also include the following processing:

[0108] Step 105: Obtain execution result data and / or network optimization effect data generated based on the network optimization processing, and adjust the parameters of the target intelligent agent based on the execution result data and / or the network optimization effect data.

[0109] Optionally, after a network controller such as an SD-WAN controller sends the forwarding table and network resource configuration commands to POP nodes and switches, the system can perform network optimization processing such as routing and security policy adjustment based on the new network configuration (new network configuration represented by the forwarding table and network resource configuration commands), while monitoring the network performance changes and security status brought about by these optimization processing, and collecting execution result data and / or network optimization effect data generated based on these network optimization processing.

[0110] Specifically, the POP nodes and switches can collect the execution results, network optimization effects and other data generated by these network optimization processes based on their data collection units, and upload the collected data to the central data processing unit. The central data processing unit can continuously iterate and optimize the target intelligent agent based on the received execution result data, network optimization effect data and other feedback information to adjust the parameters of the target intelligent agent. For example, adjust the model parameters of the AI ​​orchestrator or retrain to adapt to new network environments and challenges.

[0111] Therefore, after network optimization, the data collected by POP nodes and switches and reported to the data processing center includes not only key network data, but also execution result data and network optimization effect data generated after network optimization, to ensure that the AI ​​orchestrator can continuously iterate and optimize through these feedbacks.

[0112] Among them, the execution result data generated after network optimization processing may include but is not limited to the execution of forwarding table QoS strategy, security strategy, etc. in actual applications, such as whether the network path is successfully switched, whether the data packet forwarding is carried out according to the new path, whether the firewall rules are adjusted normally according to the security protection measures, whether specific ports are enabled or disabled normally, etc.

[0113] The network optimization effect data generated after the network optimization process may include but is not limited to network performance, network security, network real-time status and other aspects of the optimization effect data. The optimization effect data of each aspect are as follows:

[0114] 1) Network performance optimization effect data: PoP nodes monitor the performance changes of the current network, including key performance indicators such as latency, bandwidth utilization, and packet loss rate after path optimization. These data can reflect the effect of the optimized path generated by the AI ​​orchestrator in actual applications to indicate whether the network performance has been improved.

[0115] 2) Network security optimization effect data: If the optimization process involves the adjustment of security protection strategies, the PoP node will also feedback data related to security effects, such as whether the detected threats have been reduced, and changes in network traffic after the firewall strategy has been adjusted.

[0116] 3) Optimization effect data of real-time network status: After optimization, PoP nodes will continuously upload link status information, including whether the link is stable, whether new network bottlenecks have appeared, etc. This status information is the key basis for continuous optimization and dynamic adjustment of AI models.

[0117] This embodiment monitors the changes in network performance and security status caused by network optimization processing, collects execution result data and / or network optimization effect data generated based on these network optimization processing, and feeds the collected data back to the central data processing unit for continuous optimization of the target intelligent body. It can effectively improve the prediction and decision-making capabilities of the target intelligent body, and correspondingly improve the accuracy and effectiveness of network decisions, facilitate the system to dynamically adjust the network configuration, and ensure that the network is always in the best state.

[0118] In an optional embodiment, the data collection unit on the data collection device such as the POP node can realize the collection of key network data through network traffic monitoring, link performance monitoring and / or linkage with corresponding devices in the wide area network such as security devices and network controllers. Specifically, it includes but is not limited to: the data collection unit collects network traffic data of the wide area network by monitoring the traffic flowing through the POP node; collects performance index data of the wide area network by monitoring the link performance of the link connected to the POP node; and obtains security log data of the wide area network by linkage with corresponding security devices in the wide area network.

[0119] The security device is a device responsible for network security, including but not limited to a firewall.

[0120] POP nodes can obtain various key network data through the following mechanisms.

[0121] During the forwarding process, data packets will pass through one or more private clouds or public clouds. The POP nodes deployed in the public cloud or private cloud can collect / count the number of data packets sent to each WAN and the response time of the data packets, so as to collect the corresponding network data of at least one WAN connected to the POP node. The reason why the POP node can collect key network data is not simply because the data packet passes through it during the forwarding process, but also includes the linkage between the POP node and security devices such as firewalls, SD-WAN controllers, and SD-WAN path selection AI orchestrators. The specific data collection mechanism is detailed as follows:

[0122] 1) Traffic monitoring and data collection: The POP node itself has data collection units that can monitor and record all network traffic passing through the POP node in real time. The POP node can not only record the number of data packets and response time, but also capture more performance indicators (such as packet loss rate, bandwidth utilization, etc.). These data are captured through the SDN (Software Defined Network) flow table. The POP node will monitor the traffic in detail in specific flow table items according to the instructions of the controller.

[0123] 2) Performance collection across WANs: When connecting to different WANs, POP nodes will continuously monitor the links between each WAN, including but not limited to key performance indicators such as transmission delay, jitter, and bandwidth consumption. Therefore, POP nodes can not only collect data from the WAN they are connected to, but also obtain overall network performance data across WANs by monitoring multiple links.

[0124] 3) Obtaining security event logs: The security event logs can be obtained in the following two ways:

[0125] 3-1. Abnormal traffic detection: POP nodes analyze the traffic patterns passing through them in real time and can detect abnormal traffic behaviors, such as DDoS (Distributed Denial of Service attack), traffic surges or traffic chaos, etc. This AI- and rule-based traffic detection mechanism can identify potential security threats without directly parsing data packets, and record these security events in security logs.

[0126] 3-2. Linkage with security devices: POP nodes are linked with SD-WAN controllers and other security devices (such as firewalls, IDS / IPS systems). When security devices detect network attacks or potential threats, POP nodes receive notifications of these security events and record them in the security event log. This enables POP nodes to capture and process security events from the perspective of the entire network.

[0127] The POP node can obtain the performance data and security event logs of the WAN through its built-in data collection unit without directly parsing the content of each data packet. It mainly achieves comprehensive network data collection and security log recording through traffic monitoring, anomaly detection and collaboration with other security systems.

[0128] This embodiment utilizes the data collection units set on POP nodes and other devices to perform comprehensive network data collection and security log recording based on traffic monitoring, anomaly detection and collaboration with other security systems, thereby ensuring that key network data such as network traffic, performance indicators and security event logs can be captured in real time and comprehensively, thereby providing the necessary data support for subsequent dynamic intelligent network optimization decisions based on artificial intelligence, thereby ensuring that the network always operates in the best routing state, responding to network security threats in real time, automatically optimizing security policies, enhancing the network's defense capabilities, and ensuring the security and reliability of the network.

[0129] An application example of the method disclosed in this application is provided below.

[0130] In this example, the wide area network is SD-WAN, and this example implements an AI-driven SD-WAN system based on the method of the present application.

[0131] For the AI-driven SD-WAN system, the components involved, and related operations and descriptions, please see Table 1.

[0132] Table 1

[0133]

[0134] In this example, refer to Figure 5 As shown in the implementation flow chart, the implementation process of the above-mentioned AI-driven SD-WAN system from system construction to dynamic network optimization based on AI is as follows:

[0135] Step 1: Deploy PoP nodes.

[0136] Location: Multiple public and private cloud environments around the world.

[0137] Operation: Deploy PoP nodes in various public and private cloud environments to ensure network coverage and load balancing. These PoP nodes will serve as collection and forwarding points for key network data to support efficient operation and optimization of the network.

[0138] Description: PoP nodes need to have high-performance hardware configuration and reliable software system to support data collection and forwarding table execution. The network topology needs to be considered during deployment to ensure that all key network areas are covered by PoP nodes.

[0139] Step 2: Built-in data acquisition unit.

[0140] Location: In each PoP node. It can also be set in the switch.

[0141] Operation: Data acquisition units are installed in each PoP node and switch to capture data such as network traffic, performance indicators, and security event logs in real time.

[0142] Description: The data acquisition unit needs to be able to process a large amount of network data and transmit this data to the central data processing unit in real time. Among them, ensuring the integrity and security of the data is the key, and the data acquisition unit should have the characteristics of high reliability and high security.

[0143] Data collection points such as PoP nodes can also serve as data pre-processing centers to enhance the efficiency and security of data collection. The key network data collected includes but is not limited to network traffic data, performance indicators, device status, and security event logs. The system pays special attention to real-time performance parameters of the network such as bandwidth utilization, latency, packet loss rate, and security threat indicators.

[0144] Step 3: Deploy the central data processing unit.

[0145] Location: Cloud data processing center.

[0146] Operation: Deploy the SD-WAN path selection AI orchestrator and SD-WAN controller in the central data processing unit.

[0147] Description: The central data processing unit requires powerful computing and storage capabilities to process the large amount of data transmitted from PoP nodes and switches. The SD-WAN path selection AI orchestrator is responsible for data analysis and AI model training and intelligent processing after training, and the SD-WAN controller configures and manages the network based on the decisions of the AI ​​orchestrator.

[0148] Step 4: Train the AI ​​model and perform data analysis.

[0149] Location: Central Data Processing Unit.

[0150] Operation: Use the collected network data to train the AI ​​model and analyze the network status.

[0151] Description: AI orchestrators identify patterns and potential problems in the network by processing and analyzing data. Trained AI models / AI orchestrators can analyze network status in real time and generate corresponding network optimization decisions.

[0152] The data collected by PoP nodes and switches can be transmitted to the data processing center through a secure communication protocol. In this center, an efficient AI orchestrator conducts in-depth analysis of the data, identifies network traffic patterns through machine learning algorithms, and predicts potential network congestion and security threats.

[0153] Based on the insights analyzed by the AI ​​orchestrator, the decision generation unit automatically formulates and issues network optimization decisions, including but not limited to routing adjustment strategies, and network resource reconfiguration strategies such as bandwidth allocation / failover / security measures. All decisions are aimed at improving network performance, ensuring business continuity and network security.

[0154] The SD-WAN controller then configures and manages the network based on the decisions made by the AI ​​orchestrator.

[0155] Step 5: Receive decision commands and issue forwarding tables.

[0156] Location: SD-WAN controller, each PoP node, and switches.

[0157] Operation: The SD-WAN controller generates and sends forwarding table and other configuration information to each PoP node and switch.

[0158] Description: Based on the decisions made by the AI ​​orchestrator, the SD-WAN controller generates a forwarding table and related network resource configuration commands and sends them to each PoP node and switch to guide them to perform packet forwarding operations and bandwidth allocation, load balancing, failover or security policy configuration. Among them, ensuring the timely delivery and correct execution of the forwarding table is the key.

[0159] Step 6: Implementation and feedback.

[0160] Location: Each PoP node, switch and central data processing unit.

[0161] Operation: PoP nodes and switches forward packets according to the forwarding table, configure network resources according to relevant network resource configuration commands, such as bandwidth allocation, load balancing, failover or security policy configuration, and feed back the execution results and network optimization effect data generated based on the forwarding table and network resource configuration to the central data processing unit.

[0162] Description: After receiving the forwarding table, the PoP node / switch forwards the data packet according to the instructions. At the same time, it monitors the optimization effect and transmits the feedback data back to the central data processing unit for further analysis and optimization. The same is true for the optimization configuration of network resources.

[0163] Step 7: Continue to optimize and improve.

[0164] Location: Central Data Processing Unit.

[0165] Operation: Continuously optimize AI models based on feedback data to improve the accuracy and effectiveness of network decisions.

[0166] Description: The AI ​​orchestrator continuously optimizes its AI model based on feedback data to improve the model's prediction and decision-making capabilities. The system dynamically adjusts the network configuration to ensure that the network is always in the best condition.

[0167] The embodiment of the present application uses AI models such as machine learning models to analyze network traffic and performance data in real time, and can dynamically and intelligently select the best data transmission path. Compared with traditional SD-WAN, the system can more effectively process and respond to changes in network status and automatically adjust routing strategies to adapt to current network conditions.

[0168] In addition, the security of SD-WAN in known technologies mainly relies on traditional network security measures such as static firewall rules and access control lists (ACLs). Although these measures are effective, they are difficult to respond to new security threats and complex attack patterns in real time. The embodiment of the present application greatly improves the security of the network by introducing AI-driven security policy optimization. The AI ​​model can automatically identify abnormal behaviors and potential threats and dynamically adjust security policies by analyzing real-time network traffic, threat detection information, and security logs, including but not limited to:

[0169] 1) Dynamically update security policies: The AI ​​model can automatically generate new firewall rules, ACL updates and other security policies by learning the latest attack patterns. This dynamic response capability enables the network to respond to new attacks more promptly and effectively.

[0170] 2) Real-time threat detection and response: When analyzing network traffic, the AI ​​model can identify potential DDoS attacks, traffic anomalies, or malicious behavior in real time, and automatically initiate protective measures such as traffic restriction and route isolation to ensure that threats are handled at an early stage and reduce the impact on the network.

[0171] 3) Intelligent threat prevention: Through AI’s learning capabilities, the system can predict possible security risks and take preventive measures in advance. For example, based on historical network traffic data and threat intelligence, more security resources can be allocated to high-risk areas in advance.

[0172] 4) Security improvement in path optimization: In the existing SD-WAN architecture, path selection usually only considers performance factors such as bandwidth and latency, while ignoring the security of the path. This application introduces security factors into the path selection algorithm to ensure that not only performance but also path security is considered when selecting a path, including the encryption status of the traffic, the security level of the network nodes, etc. The AI ​​model will evaluate the security risk of each available path to avoid routing traffic to links or nodes that may pose threats.

[0173] 5) Security protection for access devices: In traditional SD-WAN architecture, access devices (such as PoP nodes and switches) usually only provide basic access control and identity authentication. This application uses AI to perform real-time analysis of traffic patterns and security logs of access devices, which can quickly detect possible device intrusions or unauthorized device access behaviors, and automatically adjust access control policies to protect the security of network boundaries.

[0174] Through these technological innovations, the embodiments of the present application effectively enhance the security of the network while improving the overall network performance of the SD-WAN, especially when facing complex attacks and dynamic network environments, and can provide more flexible and intelligent protection solutions.

[0175] The present application also provides a network control system, the composition structure of which is as follows: Figure 6 As shown, including at least:

[0176] A data collection unit 10 is used to collect key network data of the wide area network;

[0177] The central data processing unit 20 is used to: determine a network optimization decision for the wide area network based on the target intelligent agent and the key network data, wherein the network optimization decision is used to improve the network communication performance of the wide area network; determine network configuration information for network devices in the wide area network based on the network optimization decision; and send the network configuration information to corresponding network devices in the wide area network to control the corresponding network devices to perform network optimization processing corresponding to the network configuration information.

[0178] In an optional implementation, the data collection unit 10 is deployed on a point-of-entry device and / or a switch of the wide area network, and is used to collect at least one of network traffic data, performance indicator data, and security log data of the wide area network;

[0179] The key network data includes at least one of the network traffic data, the performance indicator data and the security log data.

[0180] In an optional embodiment, the central data processing unit 20 includes an artificial intelligence orchestrator 201 and a network controller 202;

[0181] The artificial intelligence orchestrator 201 is used to generate and output a routing adjustment strategy and / or a network resource reconfiguration strategy for the wide area network based on at least one of the network traffic data, the performance indicator data and the security log data;

[0182] The network controller 202 is configured to: generate a forwarding table matching the route adjustment strategy, and / or generate a network resource configuration command matching the network resource reconfiguration strategy; and send the forwarding table and / or the network resource configuration command to a corresponding point-of-presence device or switch in the wide area network, so as to control the corresponding point-of-presence device or switch to perform path selection and data packet forwarding based on the received forwarding table, and / or perform network resource configuration based on the received network resource configuration command;

[0183] Among them, the network configuration information of the network device in the wide area network includes the forwarding table and / or the network resource configuration command; the network resource configuration command includes at least one of the following: bandwidth allocation configuration command, quality of service policy configuration command, load balancing configuration command, failover configuration command, security policy configuration command.

[0184] In an optional embodiment, the central data processing unit 202 is also used to: obtain execution result data and / or network optimization effect data generated based on the network optimization processing, and adjust the parameters of the target intelligent agent based on the execution result data and / or network optimization effect data.

[0185] In summary, the network control method and system provided in the embodiments of the present application have at least the following technical advantages compared to the known technologies:

[0186] 1) Efficient data collection and processing: PoP nodes are distributed all over the world and can collect and upload large-scale network data in real time to ensure the timeliness and accuracy of the data. The data processing center works in conjunction with the PoP nodes to achieve efficient data transmission and processing and improve the network's response speed.

[0187] 2) Intelligent network management and optimization improve network performance: Intelligent decision-making based on the AI ​​orchestrator enables the network to automatically adjust routing strategies and resource configurations, optimizing network performance and security; the collaborative work of PoP nodes and SD-WAN controllers ensures optimal utilization of network resources, avoiding network congestion and resource waste; through intelligent routing and real-time performance optimization, the overall efficiency and response speed of the network are significantly improved, especially under high load conditions, it can maintain stable service quality.

[0188] 3) Adaptive security protection and enhanced network security: Through security policy adjustment based on AI orchestrator, network security threats can be responded to in real time and security policies can be automatically optimized, thus providing a dynamic security protection mechanism to ensure the security and reliability of the network and enhance the network's defense capabilities. AI-driven security systems can identify new threats in real time and automatically adapt, providing more advanced security protection than traditional methods.

[0189] 4) Support global coverage and unified management: By deploying PoP nodes globally, comprehensive monitoring and management of network traffic can be achieved to support the global business needs of enterprises. The SD-WAN controller can be used to uniformly manage and coordinate PoP nodes, simplifying the complexity of network management and improving management efficiency.

[0190] 5) Dynamic resource allocation: AI technology is used to predict and analyze network demand, enabling dynamic allocation of network resources. This predictive resource management helps to effectively balance loads, optimize the use of network bandwidth, and ensure the performance requirements of key applications.

[0191] 6) Continuous optimization and iterative learning: Based on the feedback mechanism between PoP nodes / switches and data processing centers, AI models can be continuously optimized and iterated to improve the accuracy and real-time performance of network decisions. It can also dynamically adapt to changes in the network environment to ensure continuous improvement in network performance and security.

[0192] 7) Cost-effectiveness and reduced operating costs: Through intelligent network management and optimization, the dependence on traditional hardware equipment is reduced, and the need for manual intervention is reduced, thereby reducing the cost of network operation and maintenance, improving resource utilization and reducing waste.

[0193] 8) Powerful cross-cloud interoperability: Optimized cloud service management and traffic scheduling capabilities enable enterprises to more flexibly utilize multi-cloud environments, accelerating the deployment and expansion of cloud services.

[0194] The above advantages not only improve the intelligence level of WAN network management, but also provide strong support for enterprises, enabling them to maintain competitiveness in a rapidly changing market environment.

[0195] It should be noted that the various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0196] For the convenience of description, the above system or device is described by dividing it into various modules or units according to its functions. Of course, when implementing the present application, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0197] It can be known from the description of the above implementation methods that those skilled in the art can clearly understand that the present application can be implemented by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present application can be essentially or the part that makes a creative contribution can be embodied in the form of a software product, which can be stored in a storage medium such as ROM / RAM, a disk, an optical disk, etc., including several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present application or some parts of the embodiments.

[0198] Finally, it should be noted that, in this article, relational terms such as first, second, third and fourth are used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0199] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A network control method, comprising: Obtain key network data of the WAN through public cloud nodes; Based on the target agent, determining a network optimization decision for the wide area network according to the key network data; The network optimization decision is used to improve the network communication performance of the wide area network; Determining network configuration information for network devices in the wide area network according to the network optimization decision; The network configuration information is sent to corresponding network devices in the wide area network to control the corresponding network devices to perform network optimization processing corresponding to the network configuration information.

2. The network control method according to claim 1, wherein obtaining key network data of the wide area network comprises: Obtaining at least one of network traffic data, performance indicator data, and security log data of the wide area network collected by various data collection units deployed in the wide area network; Wherein, each data collection unit is deployed on a point-of-entry device and / or a switch of the wide area network; the key network data includes at least one of the network traffic data, the performance indicator data and the security log data.

3. The network control method according to claim 2, wherein the point of presence device comprises a cloud access point device deployed at least on a public cloud of the wide area network, and the cloud access point device on the public cloud is used to form a network access layer of the public cloud; The obtaining of at least one of the network traffic data, performance indicator data, and security log data of the wide area network collected by each data collection unit deployed in the wide area network comprises at least one of the following: Obtaining network traffic data of the wide area network obtained by a data acquisition unit deployed on the cloud access point device by monitoring the traffic flowing through the cloud access point device; Obtaining performance indicator data of the wide area network obtained by a data acquisition unit deployed on the cloud access point device by monitoring the link performance of the link to which the cloud access point device is connected; A data collection unit deployed on the cloud access point device is obtained, and the security log data of the wide area network is obtained by linkage with the corresponding security device in the wide area network; the security device is a device responsible for network security.

4. The network control method according to claim 2, wherein the target agent-based method determines the network optimization decision for the wide area network according to the key network data, comprising: inputting at least one of the network traffic data, the performance indicator data, and the security log data into an artificial intelligence orchestrator deployed in a central data processing unit of the wide area network; Obtain a routing adjustment strategy and / or a network resource reconfiguration strategy for the wide area network generated and output by the artificial intelligence orchestrator based on input data.

5. The network control method according to claim 4, wherein the determining of network configuration information for the network device in the wide area network according to the network optimization decision comprises at least one of the following: Based on a network controller disposed in the central data processing unit, generating a forwarding table matching the routing adjustment strategy; Based on the network controller, a network resource configuration command matching the network resource reconfiguration strategy is generated; the network configuration information of the network device in the wide area network includes the forwarding table and / or the network resource configuration command; The sending of the network configuration information to a corresponding network device in the wide area network to control the corresponding network device to perform network optimization processing corresponding to the network configuration information includes: The forwarding table and / or the network resource configuration command are sent to a corresponding point-of-presence device or switch in the wide area network through the network controller, so as to control the corresponding point-of-presence device or switch to perform path selection and data packet forwarding based on the received forwarding table, and / or to perform network resource configuration based on the received network resource configuration command; The network resource configuration command includes at least one of the following: a bandwidth allocation configuration command, a quality of service policy configuration command, a load balancing configuration command, a failover configuration command, and a security policy configuration command.

6. The network control method according to claim 1, further comprising: Obtain execution result data and / or network optimization effect data generated based on the network optimization processing, and adjust the parameters of the target intelligent agent based on the execution result data and / or the network optimization effect data.

7. A network control system, comprising: Data collection unit, used to collect key network data of the wide area network; The central data processing unit is used to: determine a network optimization decision for the wide area network based on the target intelligent agent and the key network data, wherein the network optimization decision is used to improve the network communication performance of the wide area network; determine network configuration information for network devices in the wide area network based on the network optimization decision; The network configuration information is sent to corresponding network devices in the wide area network to control the corresponding network devices to perform network optimization processing corresponding to the network configuration information.

8. The network control system according to claim 7, wherein the data acquisition unit is deployed on a point-of-entry device and / or a switch of the wide area network, and is used to collect at least one of network traffic data, performance indicator data and security log data of the wide area network; The key network data includes at least one of the network traffic data, the performance indicator data and the security log data.

9. The network control system according to claim 8, wherein the central data processing unit comprises an artificial intelligence orchestrator and a network controller; The artificial intelligence orchestrator is used to generate and output a routing adjustment strategy and / or a network resource reconfiguration strategy for the wide area network based on at least one of the network traffic data, the performance indicator data and the security log data; The network controller is configured to: generate a forwarding table matching the route adjustment strategy, and / or generate a network resource configuration command matching the network resource reconfiguration strategy; and send the forwarding table and / or the network resource configuration command to a corresponding point-of-presence device or switch in the wide area network, so as to control the corresponding point-of-presence device or switch to perform path selection and data packet forwarding based on the received forwarding table, and / or perform network resource configuration based on the received network resource configuration command; in, The network configuration information of the network device in the wide area network includes the forwarding table and / or the network resource configuration command; the network resource configuration command includes at least one of the following: bandwidth allocation configuration command, quality of service policy configuration command, load balancing configuration command, failover configuration command, security policy configuration command.

10. According to the network control system according to claim 7, the central data processing unit is also used to: obtain execution result data and / or network optimization effect data generated based on the network optimization processing, and adjust the parameters of the target intelligent body based on the execution result data and / or network optimization effect data.

Images