Service scheduling method, system, equipment and medium

Abstract

The invention relates to the technical field of communication, and provides a service scheduling method, system and device and a medium, and the method comprises the steps: intercepting a network request which is sent to a physical network card by an application program and comprises a destination IP address and a service type; recording a mapping relationship among the application package name, the service type and each destination IP address; when the network request is not the first network request, analyzing a flow characteristic of a data packet corresponding to a destination IP address in the network request; inputting the service type of the network request, the destination IP address in the network request and the traffic characteristics of the destination IP address into the trained AI model, and judging whether the output of the AI model is normal or not; and when the network request is abnormal, replacing the destination IP address in the network request based on the mapping relation. According to the scheme, the AI model is used for judging the abnormality of the traffic characteristics, and then the destination IP address in the network request is replaced, so that the terminal application fault self-recovery can be realized, and the user experience is improved.

Description

Technical Field

[0001] This disclosure belongs to the field of communication technology, and in particular relates to a service scheduling method, system, device and medium. Background Technology

[0002] The application of AI in video conferencing terminals has become one of the core directions of technological development in the industry. Numerous related technologies and standards have been proposed to promote improvements in user experience and business capabilities. For example, industry standards for user experience-oriented video and multimedia service and network collaboration technologies, computing power hosts, business user profiling, and user experience quality assessment primarily focus on the application of AI in network devices such as gateways and BRAS (Broadband Remote Access Server), aiming to optimize network performance and user experience.

[0003] However, for terminal devices carrying these services, there are currently no systematic AI application patents or industry standards. Research and technological development in this field are still in their early stages, especially in how to deeply integrate AI technology with video conferencing terminals to achieve intelligent business processing and service optimization, where there is still considerable room for exploration. In the future, with continuous technological advancements, the application of AI in video conferencing terminals is expected to gradually become a key area of ​​industry standardization, providing users with a more intelligent and efficient service experience. Summary of the Invention

[0004] To address the aforementioned issues, this disclosure provides a service scheduling method, system, device, and medium that utilizes AI models for service scheduling, thereby achieving intelligent service processing.

[0005] In a first aspect, the present invention provides a service scheduling method, comprising: Intercept network requests sent by the application to the physical network card. The network requests contain the destination IP address and service type. Record the mapping relationship between application package name, service type and destination IP address; When a network request is not the first network request, use packet analysis tools to analyze the traffic characteristics of the packets corresponding to the destination IP address in the network request; The network request's service type, the destination IP address in the network request, and its traffic characteristics are input into the trained AI model to determine whether the AI ​​model's output is normal. The trained AI model is used to identify whether the traffic characteristics of the destination IP address for each service type are normal. When the traffic characteristics of the destination IP address in a network request are abnormal, the destination IP address in the network request is changed based on the above mapping relationship, and the network request with the changed destination IP address is sent to the physical network card.

[0006] Furthermore, the mapping relationship between the application package name, service type, and destination IP addresses is recorded, including recording the mapping relationship between the application package name, service type, and destination IP addresses using either of the following two methods: Method 1: Modify the mapping relationship between the application package name, service type, and destination IP address in the application package APK record; Method 2: Modify the mapping relationship between application package name, service type and destination IP address in the network interface record of the terminal system.

[0007] Furthermore, the methods also include: Mark the destination IP address in the mapping relationship that is the same as the destination IP address in the network request as an exception.

[0008] Furthermore, when the AI ​​model outputs something that is neither normal nor abnormal, a prompt message is displayed to suggest to the user whether the application quality needs to be improved. When a request to improve application quality is received, the machine learning mechanism of the AI ​​model is triggered to add the traffic characteristics of the data packets corresponding to the destination IP in the network request to the abnormal traffic characteristics; based on the mapping relationship, the destination IP address in the network request is changed, and the network request with the changed destination IP address is sent to the physical network card. When no improvement in application quality is required, the machine learning mechanism of the AI ​​model is triggered to add the traffic characteristics of the data packets corresponding to the destination IP in the network request to the normal traffic characteristics.

[0009] Furthermore, when no improvement in application quality is required, the destination IP address in the mapping relationship that is the same as the destination IP address in the network request is marked normally.

[0010] Furthermore, when the traffic characteristics of the destination IP address are abnormal, the destination IP address in the network request is changed based on the above mapping relationship, and the network request with the changed destination IP address is sent to the physical network card, including: By utilizing the service type contained in the application's network requests, it can be determined whether the destination IP address corresponding to that service type in the mapping relationship is marked as normal; If so, select a destination IP address from the list of normal destination IP addresses, change the destination IP address in the network request to the selected destination IP address, and send the network request with the changed destination IP address to the physical network card; If none is found, an untagged destination IP address is selected, and traffic characteristics are analyzed based on its historical data packets. When the traffic characteristics are determined to be normal using an AI model, the destination IP address in the network request is changed to the untagged destination IP address selected this time, and the network request with the changed destination IP address is sent to the physical network card.

[0011] Furthermore, when the network request is not the first network request, packet analysis tools are used to analyze the traffic characteristics of the data packets corresponding to the destination IP address in the network request, including: When a network request is not the first network request, use a packet analysis tool to analyze the network requests to the destination IP address, including the first network request, and the ACK packets, including the first ACK packet from the server, to obtain the latency of each response; analyze each TCP packet with the same ACK number as the first ACK packet to obtain the cumulative transmission load data, and compare the sum of the load data with the data length in the application protocol to obtain the packet loss rate.

[0012] Secondly, the present invention provides a service scheduling system, comprising: The interception unit is used to intercept network requests sent by applications to the physical network card. The network requests contain the destination IP address and service type. The recording unit is used to record the mapping relationship between the application package name, service type, and each destination IP address; The analysis unit is used to analyze the traffic characteristics of the data packets corresponding to the destination IP address in a network request when the network request is not the first network request. The judgment unit is used to input the service type of the network request, the destination IP address in the network request and its traffic characteristics into the trained AI model, and judge whether the output of the AI ​​model is normal. The trained AI model is used to identify whether the traffic characteristics of the destination IP address of each service type are normal. The result processing unit is used to change the destination IP address in the network request based on the above mapping relationship when the traffic characteristics of the destination IP address in the network request are abnormal, and then send the network request with the changed destination IP address to the physical network card.

[0013] In a second aspect, the present invention provides an electronic device, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; The processor implements the above method when executing programs stored in memory.

[0014] Fourthly, the present invention provides a computer storage medium storing a computer program, wherein the computer program is executed by a processor to perform the method described above.

[0015] Compared with the prior art, this disclosure has the following advantages: 1. By using AI models to identify anomalies in traffic characteristics, and then using the mapping relationship between application package name, service type and destination IP address, the destination IP address in the network request can be changed, enabling self-recovery of terminal application failures and improving user experience.

[0016] 2. Self-learning the characteristics of normal and abnormal traffic in the business reduces manual intervention.

[0017] 3. For business traffic characteristics that have been learned, normal destination addresses can be used first next time to improve user experience.

[0018] This solution is based on the Android system and focuses on improving business awareness capabilities. It is applicable to set-top boxes, converged terminals and other products in the video conferencing field, and can effectively enhance the intelligence level of products and user experience.

[0019] Other features and advantages of this disclosure will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the disclosure. The objects and other advantages of this disclosure may be realized and obtained by means of the structures pointed out in the description, claims and drawings. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of this disclosure or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this disclosure. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 A flowchart of a user-aware service scheduling method according to an embodiment of this disclosure is shown; Figure 2 This diagram illustrates a network request for a media playback service using packet capture data, according to an embodiment of the present disclosure. Figure 3 This diagram illustrates a network request for a media broadcast control service based on packet capture data, according to an embodiment of the present disclosure. Figure 4 This diagram illustrates a network request for fast-forwarding in a media broadcast control service, based on packet capture data, according to an embodiment of this disclosure. Figure 5A block diagram of a service scheduling system according to an embodiment of the present disclosure is shown; Figure 6 A block diagram of an electronic device according to an embodiment of the present disclosure is shown. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, and not all embodiments. Based on the embodiments of this disclosure, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this disclosure.

[0023] Before the service scheduling method of this disclosure embodiment begins, the following preparatory work is required: 1.1: Determine the traffic analysis tool: Select a traffic analysis tool such as PCAP or NDPI to analyze the packet loss rate of data packets to the destination IP.

[0024] 1.2: Determine the machine learning algorithm, such as a backpropagation (BP) neural network model; collect offline abnormal traffic features; train the BP neural network model using the collected offline abnormal traffic features to learn the collected offline abnormal traffic features, resulting in a trained BP neural network model with basic data analysis capabilities, which can be used for abnormal traffic feature identification. The input of this trained BP neural network model is the service type, destination IP, and its corresponding packet loss rate and latency; the output is whether the traffic is normal or abnormal. Load this AI model into the terminal system, such as the Android system.

[0025] 1.3: Modify the application software APK program to enable it to record the application package name, service type, and destination IP address to the shared data space when initiating a network request; or modify the terminal system network interface to enable it to record the application package name, service type, and destination IP address to the shared data space when initiating a network request.

[0026] The service scheduling method of this disclosure embodiment is as follows: Figure 1 A schematic flowchart of a service scheduling method according to an embodiment of the present disclosure is shown, such as... Figure 1 As shown, it includes the following steps: S101: Intercept network requests sent by the application to the physical network card, wherein the network requests contain the service type and the destination IP address.

[0027] A physical network interface card (NIC), also known as a network adapter, is installed on a terminal and serves as the interface connecting the terminal and the transmission medium. In related technologies, network requests sent by the application are sent to the physical NIC, which then forwards them directly to the external transmission medium.

[0028] Specifically, the application first needs to establish a connection with the internet via the phone's network connectivity function, such as Wi-Fi or mobile data. Then, the application resolves the domain name of the server it wants to access, such as https: / / www.iqiyi.com, into the server's IP address using DNS (Domain Name System). Next, the application uses the DNS-resolved IP address as the destination IP address to initiate a network request. In this disclosed solution, the network request can be an application-layer network request such as HTTP / HTTPS, RTSP / RTP, DNS, TR069, DHCP, NTP, FTP, SNMP, etc. The request includes the service type and the destination IP address. Service types include page browsing, video playback, and service control.

[0029] The aforementioned network request should have been sent by the terminal application to the external transmission medium via the physical network card. In the scheme of this embodiment, in order to perform service scheduling, the network request is intercepted before it reaches the physical network card.

[0030] For example, after the iQiyi app starts, when the user is browsing a video source page, the iQiyi app will call the system network interface to initiate a network request for web browsing to the physical network card. When the user is playing a video source, the iQiyi app will call the system network interface to initiate a network request for media playback to the physical network card. When the user performs actions such as "pause" or "fast forward / rewind," the iQiyi app will call the system network interface to initiate a network request for media playback control to the physical network card. The solution disclosed herein intercepts these network requests before they reach the physical network card.

[0031] To explain the business type in the network request, Figure 2 The image shows network requests for media playback services, captured using packet capture data: the front box is the video destination IP, and the back box is the video address. Based on this video address, it can be identified that the network request is for media playback services.

[0032] Figure 3 The image shows network requests for media broadcast control services, displayed using packet capture data: the boxed area represents the control interface, and the network requests identified from the control interface are those for media broadcast control services.

[0033] Figure 4The image shows a network request for fast-forwarding in the media broadcast control service, captured using packet capture data. The destination IP address is in the box in front of the request, and the scale in the box behind it indicates that it is 8x fast-forwarding. Therefore, the network request identified is also a network request from the media broadcast control service.

[0034] S102: Record the mapping relationship between the application package name, service type and destination IP address; In this disclosed scheme, the mapping relationship between the application package name, service type, and destination IP address can be recorded in the following two ways: Method 1: Modify the application package (APK) to record the mapping relationship between the application package name, service type, and destination IP address.

[0035] Specifically, when initiating a network request, the modified application APK records the application package name, service type, and destination IP address in a shared data space similar to a database.

[0036] Method 2: Modify the network interface of the terminal system (e.g., Android system) to record the mapping relationship between the application package name, service type and destination IP address.

[0037] Specifically, when initiating a network request, the modified network interface records the mapping relationship between the application package name, service type, and destination IP address in a shared data space similar to a database.

[0038] The aforementioned network interfaces include different application layer network protocol interfaces.

[0039] Because the current Android system lacks the functionality to record the mapping between applications and destination IP addresses, a suitable modification point in the system software is needed to add the mapping logic between applications and destination addresses. In method two, by modifying the terminal system's network interface, the corresponding service type and destination IP address are dynamically recorded based on the currently running foreground application and the network requests sent by that application.

[0040] It should be noted that when an application makes multiple requests, it can record multiple IP addresses resolved by DNS, disaster recovery addresses issued by the business platform, and manually pre-configured backup addresses. The application package (APK) records all IP addresses corresponding to each business type, uses them as destination IP addresses, and records the mapping relationship in a shared data space that can be accessed by different processes.

[0041] Considering that the same application may trigger multiple business types, and different business types use different destination IP addresses, the above-mentioned S102 of this disclosure records the mapping relationship between the application package name, business type and destination IP address.

[0042] S103: Determine whether the network request in S101 is the first network request. If yes, proceed to S106; otherwise, proceed to step S104. When the application network makes its first network request, no traffic has been generated yet, and it is impossible to analyze whether the traffic to the destination IP address is normal or abnormal. Therefore, the solution disclosed in this paper allows the first network request to pass through and sends it directly to the physical network card.

[0043] S104: Use packet analysis tools to analyze the traffic characteristics of the packets corresponding to the destination IP address in the network request. Then execute S105.

[0044] Traffic characteristics include: packet loss rate and latency; Since the message contains the time of sending the request and the time of receiving the reply, the delay can be obtained; since the message contains the theoretical number of data bytes that should be sent and the number of data bytes that the other party actually received, the packet loss rate can be obtained.

[0045] In step 104, DPI (Deep Packet Inspection) or PCAP (Packet Capture) technology can be used to analyze the packet loss rate and latency of application layer network protocols such as HTTP / HTTPS, RTSP / RTP, DNS, TR069, DHCP, NTP, FTP, and SNMP corresponding to the application's IP address.

[0046] Specifically, based on open-source packet analysis tools such as PCAP, NDPI, and Wireshark, the function of identifying the ACK status of the TCP protocol for each destination IP address to be analyzed is added.

[0047] When an application initiates a network request, a data analysis tool is triggered to analyze the network requests, including the first network request, and the ACK packets, including the first ACK packet from the server, for the destination IP address to be analyzed. The response time of each response is obtained, which is also known as latency. The TCP packets with the same ACK number as the first ACK packet are analyzed one by one to obtain the cumulative transmission load data. The sum of the load data is compared with the data length in the application protocol (HTTP Content-Length) to obtain the packet loss rate.

[0048] S105: Input the service type of the network request, the destination IP address in the network request and its traffic characteristics into the trained AI model, and determine whether the output of the AI ​​model is normal; if yes, proceed to S106; if no, proceed to step 109. The basic principle behind the AI ​​model's judgment after training is as follows: Based on the business type, identify whether the packet loss rate and latency meet the characteristics of normal traffic. If the current business type is web browsing, focus on the packet loss rate; if the current business type is video playback or control-related business, focus on both packet loss rate and latency.

[0049] Different types of services have different network requirements: web browsing has the lowest requirements, only needing no packet loss. Media playback requires higher bandwidth. Media playback control requires low latency. Therefore, the characteristics for analyzing the quality of different services will also differ.

[0050] After inputting the service type of the network request, the destination IP address in the network request, and its traffic characteristics into the trained AI model, the AI ​​model outputs whether the traffic characteristics corresponding to the destination IP address are normal, abnormal, or uncertain.

[0051] S106: Send the network request to the physical network card, which then sends the network request to the transmission medium outside the terminal, enabling the application to send network requests to the outside. Then execute S107. S107: Mark the destination IP address in the network request as normal and record it in the shared data space, then execute S108; S108: Monitor data packets destined for IP addresses.

[0052] The data packets monitored in step S108 are used in the next round in step 104.

[0053] S109: Determine whether the AI ​​model output is abnormal. If yes, execute S110; otherwise, execute S115. Step 109 is executed when the AI ​​model outputs a judgment result that is not normal in step 105 above. When the judgment result of step 109 is negative, it means that the AI ​​model output is neither normal nor abnormal. Since the AI ​​model output is in the uncertain state among the three states of normal, abnormal or uncertain, when the judgment result of step 109 is negative, the AI ​​model output is uncertain.

[0054] S110: Mark the destination IP address in the network request as abnormal and record it in the shared data space, then execute S111; S111: Based on the mapping relationship between application installation package, service type, and IP address recorded in the shared data space, determine whether the IP address corresponding to the service type carried in the network request of the application exists and is marked as normal; if it exists, proceed to S112; if it does not exist, proceed to S118. S112: Select a destination IP address from the destination IP addresses marked as normal, modify the destination IP address in the network request to the selected destination IP address, obtain the modified network request, and then execute S113; S113: Send the modified network request to the physical network card, and then execute S114.

[0055] Here, the physical network card sends the modified network request to the transmission medium outside the terminal, enabling the application to send network requests outward.

[0056] S114: Monitor packets to the selected IP address.

[0057] S115: Output a prompt message to the user asking if they need to improve the application quality; if yes is received, execute S116; if no is received, execute S117. S116: Trigger the AI ​​model's learning mechanism: Add the analyzed traffic characteristics of the destination IP address to the abnormal traffic characteristic database corresponding to that destination IP address, and then execute S111; S117: Trigger the AI ​​model's learning mechanism: Add the traffic characteristics of the destination IP address in the network request to the normal traffic characteristic library corresponding to that destination IP address, and then execute S106; S118: Select an unlabeled destination IP address, utilize the historical data packets returned by the server corresponding to the unlabeled destination IP address, and combine them with the service type of the network request to analyze the traffic characteristics of the unlabeled IP address; then execute S105.

[0058] Considering that users are most concerned about the experience of applications in the foreground, preferably, after S102 and before S104, the method further includes: Based on the interfaces of the foreground applications already obtained by the Android system, and the mapping relationship between applications and IPs recorded in steps S101 and S102, determine whether the application corresponding to the current IP is in the foreground. If it is in the foreground, execute step S104. If it is in the background, skip step S104 and directly send a network request.

[0059] Based on the above method, this disclosure also provides a service scheduling system corresponding to the above method. Figure 5 A block diagram of a service scheduling system according to an embodiment of the present disclosure is shown, such as Figure 5 As shown, the service scheduling system includes: an interception unit, a recording unit, an analysis unit, a judgment unit, and a judgment result processing unit, wherein: Interception unit 51 is used to intercept network requests sent by applications to the physical network card. The network requests contain the destination IP address and service type. Recording unit 52 is used to record the mapping relationship between application package name, service type and destination IP address; Analysis unit 53 is used to analyze the traffic characteristics of the data packets corresponding to the destination IP address in the network request when the network request is not the first network request; The judgment unit 54 is used to input the service type of the network request, the destination IP address in the network request and its traffic characteristics into the trained AI model, and judge whether the output of the AI ​​model is normal. The trained AI model is used to identify whether the traffic characteristics of the destination IP address of each service type are normal. The result processing unit 55 is used to change the destination IP address in the network request based on the above mapping relationship when the traffic characteristics of the destination IP address in the network request are abnormal, and then send the network request with the changed destination IP address to the physical network card.

[0060] Furthermore, the recording unit 52 is specifically used to record the mapping relationship between the application package name, service type, and destination IP addresses in either of the following two ways: Method 1: Modify the mapping relationship between the application package name, service type, and destination IP address in the application package APK record; Method 2: Modify the mapping relationship between application package name, service type and destination IP address in the network interface record of the terminal system.

[0061] Furthermore, the judgment result processing unit is also used to mark the destination IP address in the mapping relationship that is the same as the destination IP address in the network request as an anomaly.

[0062] Furthermore, the judgment result processing unit 55 is also used to output a prompt message when the AI ​​model output is neither normal nor abnormal, prompting the user whether application quality needs to be improved; when it receives a request to improve application quality, it triggers the machine learning mechanism of the AI ​​model to add the traffic characteristics of the data packet corresponding to the destination IP in the network request to the abnormal traffic characteristics; based on the mapping relationship, it changes the destination IP address in the network request and sends the network request with the changed destination IP address to the physical network card; when it receives a request that application quality does not need to be improved, it triggers the machine learning mechanism of the AI ​​model to add the traffic characteristics of the data packet corresponding to the destination IP in the network request to the normal traffic characteristics.

[0063] Furthermore, the judgment result processing unit 55 is also used to mark the destination IP address in the mapping relationship that is the same as the destination IP address in the network request as normal when it receives a result that does not require improvement of application quality.

[0064] Furthermore, the judgment result processing unit 55 is specifically used to determine whether the destination IP address corresponding to the service type in the mapping relationship is marked as normal based on the service type contained in the application's network request; if so, select a destination IP address from the marked normal destination IP addresses, replace the destination IP address in the network request with the selected destination IP address, and send the network request with the changed destination IP address to the physical network card; if not, select an unmarked destination IP address, perform traffic feature analysis based on its historical data packets, and when the traffic feature is determined to be normal using an AI model, replace the destination IP address in the network request with the unmarked destination IP address selected this time, and send the network request with the changed destination IP address to the physical network card.

[0065] Furthermore, the analysis unit 53 is specifically used to analyze, when the network request is not the first network request, the network requests to the destination IP address, including the first network request, and the ACK packets, including the first ACK packet from the server, to obtain the latency of each reply; analyze the TCP packets with the same ACK number as the first ACK packet one by one to obtain the cumulative transmission load data, and compare the sum of the load data with the data length in the application protocol to obtain the packet loss rate.

[0066] Based on the same inventive concept as the above-disclosed content, this disclosure also provides an electronic device, the structural block diagram of which is shown below. Figure 6 As shown. An electronic device according to an embodiment of this disclosure includes at least one processor and at least one memory electrically connected to the processor, wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method described above.

[0067] It should be noted that the electrical connection between the above-mentioned units does not necessarily mean the connection between lines. The indirect connection method can be applied to the embodiments of this disclosure as long as it achieves the purpose of this disclosure.

[0068] Based on the same inventive concept, this disclosure also provides a computer storage medium storing a computer program, which, when executed by a processor, implements the steps of the above method.

[0069] Although the present disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present disclosure.

Claims

1. A service scheduling method, characterized in that, include: Intercept network requests sent by the application to the physical network card, the network requests containing the destination IP address and service type; Record the mapping relationship between application package name, service type and destination IP address; When a network request is not the first network request, use packet analysis tools to analyze the traffic characteristics of the packets corresponding to the destination IP address in the network request; The network request's service type, the destination IP address in the network request, and its traffic characteristics are input into the trained AI model to determine whether the AI ​​model's output is normal. The trained AI model is used to identify whether the traffic characteristics of the destination IP address for each service type are normal. When the traffic characteristics of the destination IP address in a network request are abnormal, the destination IP address in the network request is changed based on the above mapping relationship, and the network request with the changed destination IP address is sent to the physical network card.

2. The method according to claim 1, characterized in that, Record the mapping relationship between the application package name, service type, and destination IP addresses, including: recording the mapping relationship between the application package name, service type, and destination IP addresses using any of the following two methods: Method 1: Modify the mapping relationship between the application package name, service type, and destination IP address in the application package APK record; Method 2: Modify the mapping relationship between application package name, service type and destination IP address in the network interface record of the terminal system.

3. The method according to claim 1, characterized in that, The method further includes: Mark the destination IP address in the mapping relationship that is the same as the destination IP address in the network request as an exception.

4. The method according to any one of claims 1-3, characterized in that, When the AI ​​model outputs something that is neither normal nor abnormal, a prompt message is displayed to suggest to the user whether the application quality needs to be improved. When a need to improve application quality is received, the machine learning mechanism of the AI ​​model is triggered to add the traffic characteristics of the data packets corresponding to the destination IP in the network request to the abnormal traffic characteristics; Based on the mapping relationship, the destination IP address in the network request is changed, and the network request with the changed destination IP address is sent to the physical network card. When no improvement in application quality is required, the machine learning mechanism of the AI ​​model is triggered to add the traffic characteristics of the data packets corresponding to the destination IP in the network request to the normal traffic characteristics.

5. The method according to claim 4, characterized in that, When no improvement in application quality is required, the destination IP address in the mapping relationship that is the same as the destination IP address in the network request is marked normally.

6. The method according to any one of claims 1-3, characterized in that, When the traffic characteristics of the destination IP address in a network request are abnormal, the destination IP address in the network request is changed based on the above mapping relationship, and the network request with the changed destination IP address is sent to the physical network card, including: By utilizing the service type contained in the application's network requests, it can be determined whether the destination IP address corresponding to that service type in the mapping relationship is marked as normal; If so, select a destination IP address from the list of normal destination IP addresses, change the destination IP address in the network request to the selected destination IP address, and send the network request with the changed destination IP address to the physical network card; If none is found, an untagged destination IP address is selected, and traffic characteristics are analyzed based on its historical data packets. When the traffic characteristics are determined to be normal using an AI model, the destination IP address in the network request is changed to the untagged destination IP address selected this time, and the network request with the changed destination IP address is sent to the physical network card.

7. The method according to any one of claims 1-3, characterized in that, When a network request is not the first network request, use packet analysis tools to analyze the traffic characteristics of the packets corresponding to the destination IP address in the network request, including: When a network request is not the first network request, use a packet analysis tool to analyze the network requests to the destination IP address, including the first network request, and the ACK packets, including the first ACK packet from the server, to obtain the latency of each response; analyze each TCP packet with the same ACK number as the first ACK packet to obtain the cumulative transmission load data, and compare the sum of the load data with the data length in the application protocol to obtain the packet loss rate.

8. A service scheduling system, characterized in that, include: The interception unit is used to intercept network requests sent by applications to the physical network card, wherein the network requests contain the destination IP address and service type; The recording unit is used to record the mapping relationship between the application package name, service type, and each destination IP address; The analysis unit is used to analyze the traffic characteristics of the data packets corresponding to the destination IP address in a network request when the network request is not the first network request. The judgment unit is used to input the service type of the network request, the destination IP address in the network request and its traffic characteristics into the trained AI model, and judge whether the output of the AI ​​model is normal. The trained AI model is used to identify whether the traffic characteristics of the destination IP address of each service type are normal. The result processing unit is used to change the destination IP address in the network request based on the above mapping relationship when the traffic characteristics of the destination IP address in the network request are abnormal, and then send the network request with the changed destination IP address to the physical network card.

9. An electronic device, characterized in that, It includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; A processor, when executing a program stored in memory, implements the method described in any one of claims 1-7.

10. A computer storage medium, characterized in that, The computer storage medium stores a computer program, which, when executed by a processor, implements the method described in any one of claims 1-7.

Images