A content delivery network configuration method, system, and apparatus based on SVM

Abstract

This invention relates to the field of content delivery network (CDN) technology, and discloses a CDN configuration method, system, and apparatus based on SVM. The method includes: receiving alarm information and a target outage rate from a central cluster; obtaining initial static configuration data and dynamic data for all domains in the region where the edge cluster is located; determining a parameter tuning step size based on the initial static configuration data; obtaining target data based on the initial static configuration data, dynamic data, and parameter tuning step size; updating a pre-trained SVM algorithm model based on the target data to obtain a target SVM algorithm model; and using the target SVM algorithm model, with the target outage rate as the objective, determining the target static configuration data and sending it to the central cluster. This invention improves response speed and stability, enhances CDN performance, and improves user experience and satisfaction by promptly receiving alarm information, prioritizing user detection of outage issues, and providing personalized configurations for domains experiencing anomalies.

Description

Technical Field

[0001] This invention relates to the field of content delivery network technology, and specifically to a content delivery network configuration method, system, and apparatus based on SVM. Background Technology

[0002] With the rapid development of internet technology, Content Delivery Networks (CDNs) have become a critical infrastructure for ensuring users can access websites and applications efficiently and stably. However, in practical applications, CDNs face many challenges, especially in handling user reports and optimizing configurations.

[0003] Currently, common user-reported failure scenarios in CDN services include video playback interruptions and webpage access failures. Existing solutions typically involve problem analysis and configuration adjustments after a user reports a failure. This reactive approach not only results in long response times but also negatively impacts user experience. Furthermore, due to significant differences in network environments and user behavior across different regions, a uniform CDN configuration for all regions is insufficient to meet the personalized needs of each area. Therefore, existing CDN configuration methods suffer from long response times, poor user experience, and an inability to meet personalized requirements. Summary of the Invention

[0004] In view of this, the present invention provides a content delivery network configuration method, system and device based on SVM to solve the problems of long response time, impact on user experience and inability to meet personalized needs in existing CDN configuration methods.

[0005] In a first aspect, the present invention provides an SVM-based content delivery network configuration method, applied to the edge cluster of a content delivery network, the method comprising:

[0006] Receive alarm information and target outage rate sent by the central cluster of the content delivery network, obtain the initial static configuration data and dynamic data of all domain names in the region where the edge cluster is located, and the alarm information indicates that the domain name of the content delivery network is abnormal;

[0007] Based on the initial static configuration data of all domains, determine the parameter tuning step size;

[0008] Based on the initial static configuration data, dynamic data, and parameter tuning step size, the target data is obtained;

[0009] The pre-trained SVM algorithm model is updated based on the target data to obtain the target SVM algorithm model;

[0010] The target SVM algorithm model is adopted, with the target interruption rate as the objective, to determine the target static configuration data, and then send the target static configuration data to the central cluster so that the central cluster can optimize the configuration according to the target static configuration data.

[0011] The content delivery network configuration method based on SVM provided in this invention allows the edge cluster to quickly respond to domain name anomalies by receiving alarm information. It then acquires initial static configuration data and dynamic data, and determines the parameter tuning step size based on the initial static configuration data. This determination of the parameter tuning step size makes the optimization process more refined, flexibly adapting to various situations. Combining the static configuration data, dynamic data, and parameter tuning step size forms target data, providing comprehensive input for subsequent model updates. By updating the SVM algorithm model, it can adapt to new data distributions, improving the model's generalization ability on new data and coping with constantly changing network environments. Since the target SVM algorithm model has learned the relationship between static configuration data and the outage rate, when the target outage rate is the objective, the target SVM algorithm model can accurately determine the target static configuration data through parameter tuning. By receiving alarm information in a timely manner, users are prioritized to discover outage problems, thereby enabling personalized configuration for domain names experiencing anomalies, improving response speed and stability, enhancing the performance of the content delivery network, reducing the outage rate, and improving user experience and satisfaction.

[0012] In one optional implementation, the initial static configuration data for each domain name includes multiple static configuration parameters and their corresponding data;

[0013] Based on the initial static configuration data of all domains, determine the parameter tuning step size, including:

[0014] Based on the initial static configuration data corresponding to all domain names, for each static configuration parameter in the initial static configuration data, determine the average value of the static configuration parameters;

[0015] Determine the standard deviation of the static configuration parameters based on their average value.

[0016] Based on the normal distribution and the standard deviation of the static configuration parameters, the tuning step size of the static configuration parameters is determined.

[0017] The content delivery network configuration method based on SVM provided in this invention calculates the average value and standard deviation of each static configuration parameter in the initial static configuration data. Based on the normal distribution and the standard deviation, the tuning step size of each static configuration parameter can be determined more accurately. A reasonable tuning step size helps to maintain the stability of the model and avoid model performance fluctuations caused by improper step size.

[0018] In one optional implementation, before updating the pre-trained SVM algorithm model based on the target data to obtain the target SVM algorithm model, the method further includes:

[0019] Acquire training static data and sample dynamic data, including the interruption rate;

[0020] The average value of the data in the sample dynamic data, excluding the interruption rate, is calculated according to time and domain name to obtain the training dynamic data;

[0021] Using the SVM algorithm model, predictions are made based on training static data and training dynamic data to obtain multiple training interruption rates;

[0022] Determine multiple training errors between multiple training interruption rates and the interruption rates in the training dynamic data;

[0023] The SVM algorithm model is trained based on multiple training errors to obtain a pre-trained SVM algorithm model.

[0024] The content delivery network configuration method based on SVM provided in this invention obtains training dynamic data by calculating the average value of sample dynamic data according to time and domain name. This smooths out short-term fluctuations and makes the training dynamic data more stable. The SVM algorithm model is used to predict based on the training static data and the training dynamic data to obtain the training interruption rate. This determines the training error between the training interruption rate and the actual interruption rate in the training dynamic data, thereby quantifying the prediction error of the model. By training based on the training error, the SVM algorithm model can be optimized, improving the prediction accuracy and generalization ability of the model.

[0025] In one alternative implementation, the target data includes the interruption rate;

[0026] The pre-trained SVM algorithm model is updated based on the target data to obtain the target SVM algorithm model, including:

[0027] A pre-trained SVM algorithm model is used to predict the target data, resulting in multiple first prediction interruption rates;

[0028] Determine multiple prediction errors between multiple first-predicted outage rates and the outage rates in the target data;

[0029] The pre-trained SVM algorithm model is updated based on multiple prediction errors to obtain the target SVM algorithm model.

[0030] The content delivery network configuration method based on SVM provided in this invention uses a pre-trained SVM algorithm model to predict target data, obtaining multiple first predicted interruption rates. This determines the prediction error between the first predicted interruption rate and the actual interruption rate in the target data. Based on this prediction error, the SVM algorithm model is updated to obtain the target SVM algorithm model. This allows the model to learn new data features and deeply learn the relationship between static configuration data and interruption rates, thereby improving the model's prediction accuracy and generalization ability.

[0031] In one optional implementation, a target SVM algorithm model is used to determine target static configuration data with a target interruption rate as the objective, including:

[0032] Select initial target data from the target data;

[0033] The target SVM algorithm model is used to predict the initial target data to obtain the second prediction interruption rate;

[0034] If the difference between the second predicted interruption rate and the target interruption rate is greater than the preset error threshold, the static configuration data in the initial target data is adjusted according to the parameter tuning step size, and the prediction process is repeated until the difference between the second predicted interruption rate and the target interruption rate is no greater than the preset error threshold. The static configuration data in the target data obtained from the last adjustment is then used as the target static configuration data.

[0035] The content delivery network configuration method based on SVM provided in this invention provides a clear starting point for subsequent prediction and adjustment by selecting initial target data, ensuring that the optimization process proceeds in an orderly manner. Then, the target SVM algorithm model is used to predict the initial target data to obtain a second predicted interruption rate. If the difference between the second predicted interruption rate and the target interruption rate is greater than a preset error threshold, the static configuration data in the initial target data is adjusted according to the parameter tuning step size. By gradually adjusting the static configuration data, the prediction results of the model are gradually optimized to ensure that the target interruption rate is finally achieved, thus realizing targeted configuration optimization for domains that have abnormal situations.

[0036] Secondly, this invention provides an SVM-based content delivery network configuration method, applied to the central cluster of a content delivery network, the method comprising:

[0037] Monitor the outage rate of all domains in the target region;

[0038] For any domain name, if the domain name's interruption rate exceeds a preset interruption rate threshold, an alarm message and the target interruption rate corresponding to the domain name are sent to the edge cluster of the content delivery network.

[0039] The system receives target static configuration data for the edge cluster of the content delivery network, optimizes the configuration based on the target static configuration data, and determines the target static configuration data by adopting the target SVM algorithm model with the target interruption rate as the objective. The target SVM algorithm model is obtained by updating the pre-trained SVM algorithm model based on the target data obtained by static configuration data, dynamic data, and parameter tuning step size.

[0040] The content delivery network configuration method based on SVM provided in this embodiment of the invention monitors the interruption rate in real time. When the interruption rate of any domain name exceeds the threshold, an alarm message is promptly sent to the edge cluster. Priority is given to users who discover interruption problems, and target static configuration data sent by the edge cluster is received. Based on this data, configuration optimization is performed to resolve abnormal domain name situations and improve network stability and reliability.

[0041] In one alternative implementation, the method further includes:

[0042] Verify whether the actual static configuration of the domain name matches the target static configuration data;

[0043] If the actual static configuration of a domain name is inconsistent with the target static configuration data, the domain name's static configuration will be automatically repaired.

[0044] The content delivery network configuration method based on SVM provided in this invention verifies whether the actual static configuration of domains with abnormal outage rates is consistent with the target static configuration data after configuration optimization. If there is a discrepancy, automatic repair is performed to ensure that the actual static configuration is completely matched with the optimized target configuration, thereby avoiding problems caused by configuration errors and improving the reliability and stability of the network.

[0045] Thirdly, the present invention provides an SVM-based content delivery network configuration system, the system comprising:

[0046] Monitor the outage rate of all domains in the target region;

[0047] For any domain name, if the domain name's interruption rate exceeds a preset interruption rate threshold, an alarm message and the target interruption rate corresponding to the domain name are sent to the edge cluster of the content delivery network.

[0048] Receive alarm information and target outage rate sent by the central cluster of the content delivery network, obtain the initial static configuration data and dynamic data of all domain names in the region where the edge cluster is located, and the alarm information indicates that the domain name of the content delivery network is abnormal;

[0049] Based on the initial static configuration data of all domains, determine the parameter tuning step size;

[0050] Based on the initial static configuration data, dynamic data, and parameter tuning step size, the target data is obtained;

[0051] The pre-trained SVM algorithm model is updated based on the target data to obtain the target SVM algorithm model;

[0052] The target SVM algorithm model is adopted, with the target interruption rate as the objective, to determine the target static configuration data, and then send the target static configuration data to the central cluster so that the central cluster can optimize the configuration according to the target static configuration data.

[0053] Receive target static configuration data from the edge cluster of the content delivery network, and optimize the configuration based on the target static configuration data.

[0054] Fourthly, the present invention provides an SVM-based content delivery network configuration device for use in edge clusters of content delivery networks, the device comprising:

[0055] The receiving module is used to receive alarm information and target outage rate sent by the central cluster of the content delivery network. The alarm information indicates that the domain name of the content delivery network is abnormal.

[0056] The data acquisition module is used to obtain the initial static configuration data and dynamic data of all domain names in the region where the edge cluster is located;

[0057] The algorithm module is used to determine the parameter tuning step size based on the initial static configuration data of all domain names, obtain the target data based on the initial static configuration data, dynamic data, and parameter tuning step size, update the pre-trained SVM algorithm model based on the target data to obtain the target SVM algorithm model, and use the target SVM algorithm model to determine the target static configuration data with the target interruption rate as the objective.

[0058] The upload module is used to send the target static configuration data to the central cluster so that the central cluster can optimize the configuration according to the target static configuration data.

[0059] Fifthly, the present invention provides an SVM-based content delivery network configuration device, applied to the central cluster of a content delivery network, the device comprising:

[0060] A monitoring platform used to monitor the outage rate of all domains in the target area;

[0061] The decision platform is used to send alarm information and the target interruption rate corresponding to the domain name to the edge cluster of the content delivery network when the interruption rate of the domain name exceeds the preset interruption rate threshold. It receives the target static configuration data of the edge cluster of the content delivery network, and performs configuration optimization based on the target static configuration data. The target static configuration data is determined by using the target SVM algorithm model with the target interruption rate as the objective. The target SVM algorithm model is obtained by updating the pre-trained SVM algorithm model based on the target data obtained by static configuration data, dynamic data, and parameter tuning step size.

[0062] In a sixth aspect, the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the computer instructions to perform the SVM-based content delivery network configuration method of the first aspect or any corresponding embodiment thereof, or to perform the SVM-based content delivery network configuration method of the second aspect or any corresponding embodiment thereof.

[0063] In a seventh aspect, the present invention provides a computer-readable storage medium storing computer instructions, the computer instructions being configured to cause a computer to execute the SVM-based content delivery network configuration method of the first aspect or any corresponding embodiment thereof, or to execute the SVM-based content delivery network configuration method of the second aspect or any corresponding embodiment thereof.

[0064] Eighthly, the present invention provides a computer program product, including computer instructions, which are used to cause a computer to execute the SVM-based content delivery network configuration method of the first aspect or any corresponding embodiment thereof, or to execute the SVM-based content delivery network configuration method of the second aspect or any corresponding embodiment thereof. Attached Figure Description

[0065] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0066] Figure 1 This is a schematic diagram of an SVM-based content delivery network configuration device applied to an edge cluster of a content delivery network according to an embodiment of the present invention;

[0067] Figure 2This is a flowchart of an SVM-based content delivery network configuration method for an edge cluster of a content delivery network according to an embodiment of the present invention;

[0068] Figure 3 This is a schematic diagram of an SVM-based content delivery network configuration device applied to a central cluster of a content delivery network according to an embodiment of the present invention;

[0069] Figure 4 This is a flowchart of an SVM-based content delivery network configuration method applied to a central cluster of a content delivery network according to an embodiment of the present invention;

[0070] Figure 5 This is a flowchart of an SVM-based content delivery network configuration system according to an embodiment of the present invention;

[0071] Figure 6 This is a schematic diagram of the SVM algorithm model according to an embodiment of the present invention;

[0072] Figure 7 This is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. Detailed Implementation

[0073] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0074] Common user-reported failure scenarios in CDN services include video playback interruptions and webpage access failures. Existing solutions typically analyze and adjust configurations after a user reports a failure. This reactive approach not only has a long response time but also impacts user experience. Furthermore, due to significant differences in network environments and user behavior across different regions, a uniform CDN configuration for all regions is insufficient to meet the personalized needs of each region. Therefore, existing CDN configuration methods suffer from long response times, poor user experience, and an inability to meet personalized requirements. The SVM-based content delivery network configuration method provided in this invention, by promptly receiving alarm information and prioritizing user-reported interruption issues, allows for personalized configuration of domains experiencing anomalies, improving response speed and stability. This enhances the performance of the content delivery network, reduces interruption rates, and improves user experience and satisfaction.

[0075] According to an embodiment of the present invention, an embodiment of a content delivery network configuration method based on SVM is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0076] Figure 1 This is a schematic diagram of an SVM-based content delivery network configuration device applied to an edge cluster of a content delivery network according to an embodiment of the present invention, as shown below. Figure 1 As shown, the receiving module receives alarm information and target outage rate from the central cluster of the content delivery network. The alarm information indicates that the domain name of the content delivery network is abnormal. The acquisition module acquires the initial static configuration data and dynamic data of all domain names in the area where the edge cluster is located. The algorithm module determines the parameter tuning step size based on the initial static configuration data of all domain names, obtains the target data based on the initial static configuration data, dynamic data, and parameter tuning step size, updates the pre-trained SVM algorithm model based on the target data to obtain the target SVM algorithm model, and uses the target SVM algorithm model with the target outage rate as the objective to determine the target static configuration data. The uploading module sends the target static configuration data to the central cluster so that the central cluster can optimize its configuration according to the target static configuration data.

[0077] Specifically, the edge cluster includes a receiving module, a data acquisition module, an algorithm module, and an uploading module. The receiving module receives information and data sent by the central cluster. The data acquisition module obtains the initial static configuration data and dynamic data of all domains in the region where the edge cluster is located at the current time. The algorithm module maintains an SVM algorithm model. A pre-trained SVM algorithm model is obtained by training this model. When the data acquisition module collects data based on alarm information, it updates the pre-trained model based on the latest data in the region where the edge cluster is located, obtaining the target SVM algorithm model, thereby determining the corresponding target static configuration data for the target outage rate. The uploading module sends the target static configuration data determined by the algorithm module to the central cluster so that the central cluster can perform the corresponding configuration.

[0078] This embodiment provides an SVM-based content delivery network configuration method, which can be used in edge clusters of content delivery networks. Figure 2 This is a flowchart of an SVM-based content delivery network configuration method applied to an edge cluster of a content delivery network according to an embodiment of the present invention, such as... Figure 2 As shown, the process includes the following steps:

[0079] Step S201: Receive alarm information and target outage rate from the central cluster of the content delivery network (CDN). Obtain the initial static configuration data and dynamic data of all domain names in the region where the edge cluster is located. The alarm information indicates that the domain names of the content delivery network are abnormal. Specifically, the CDN network includes a central cluster and edge clusters. The central cluster is the core part of the CDN network, responsible for global management and coordination. It is usually located in a data center and has powerful computing and storage resources. The edge cluster is the front-end part of the CDN network, distributed in geographical locations close to users. It is usually composed of multiple edge nodes, each responsible for user requests in a specific region. When the central cluster monitors any region, it monitors all domain names included in that region. If an abnormal outage rate is detected for any domain name, assuming that the service types in a region are consistent, the central cluster can randomly select an edge cluster in that region and send alarm information and the target outage rate corresponding to the domain name to that edge cluster. This allows the edge cluster to help optimize the CDN network configuration to achieve the target outage rate, thereby resolving the abnormal situation of that domain name. Compared to related technologies that use a uniform configuration across all regions, this invention presets a corresponding target outage rate for each domain name in each region. This allows for personalized configuration for each region or domain name, and alarm information takes precedence over user reports, enabling timely handling of anomalies and improving user experience. Optionally, the initial static configuration data includes the domain name's TCP (Transmission Control Protocol) packet sending strategy, timeout settings, and buffer-related configurations. These data are adjustable by the CDN network itself, meaning the outage rate can be changed by adjusting these data. The dynamic data includes the number of retransmissions for each TCP connection, the outage rate, the request method, and the request and response header sizes for each user request. These data are user-driven and cannot be adjusted. Therefore, static configuration data needs to be determined for configuration optimization.

[0080] Step S202: Based on the initial static configuration data of all domain names, determine the parameter tuning step size. Specifically, analyze the initial static configuration data to determine an appropriate parameter tuning step size, which is used to gradually adjust parameters during the optimization process to approach the target interruption rate. A reasonable parameter tuning step size can ensure more accurate parameter adjustments during the optimization process and avoid instability caused by over-adjustment.

[0081] Step S203: Based on the initial static configuration data, dynamic data, and parameter tuning step size, the target data is obtained. Specifically, the initial static configuration data and dynamic data are combined, and the initial static configuration data is gradually adjusted using the parameter tuning step size to generate multiple sets of different data configurations as target data, providing diverse samples for model training.

[0082] Step S204 involves updating the pre-trained SVM algorithm model based on the target data to obtain the target SVM algorithm model. Specifically, this pre-trained SVM algorithm model is obtained by training an existing SVM algorithm model, which already possesses the ability to accurately predict the outage rate and has learned the relationship between static configuration data and the outage rate. Through the algorithm module of the edge cluster, the pre-trained SVM algorithm model is retrained using the target data to update the model parameters, enabling the model to continuously learn new data features and enhance prediction accuracy and generalization ability.

[0083] Step S205: Using a target SVM algorithm model and a target outage rate as the objective, target static configuration data is determined and sent to the central cluster. This allows the central cluster to optimize its configuration according to the target static configuration data. Specifically, the target SVM algorithm model predicts the target data. By gradually adjusting the static configuration data, the model's prediction results are progressively optimized to ensure the target outage rate is ultimately achieved. The static configuration data at this point is then used as the target static configuration data. This target static configuration data is sent to the central cluster via the edge cluster's upload module, enabling the central cluster to make corresponding configuration adjustments. This ensures that domains experiencing abnormal situations reach the target outage rate, restoring a normal network environment and improving user experience.

[0084] The content delivery network configuration method based on SVM provided in this invention allows the edge cluster to quickly respond to domain name anomalies by receiving alarm information. It then acquires initial static configuration data and dynamic data, and determines the parameter tuning step size based on the initial static configuration data. This determination of the parameter tuning step size makes the optimization process more refined, flexibly adapting to various situations. Combining the static configuration data, dynamic data, and parameter tuning step size forms target data, providing comprehensive input for subsequent model updates. By updating the SVM algorithm model, it can adapt to new data distributions, improving the model's generalization ability on new data and coping with constantly changing network environments. Since the target SVM algorithm model has learned the relationship between static configuration data and the outage rate, when the target outage rate is the objective, the target SVM algorithm model can accurately determine the target static configuration data through parameter tuning. By receiving alarm information in a timely manner, users are prioritized to discover outage problems, thereby enabling personalized configuration for domain names experiencing anomalies, improving response speed and stability, enhancing the performance of the content delivery network, reducing the outage rate, and improving user experience and satisfaction.

[0085] Figure 3 This is a schematic diagram of an SVM-based content delivery network configuration device applied to a central cluster of a content delivery network according to an embodiment of the present invention, as shown below. Figure 3As shown, the monitoring platform is used to monitor the outage rate of all domains in the target area; the decision platform is used to send alarm information and the target outage rate corresponding to the domain to the edge cluster of the content delivery network when the outage rate of any domain exceeds the preset outage rate threshold, receive the target static configuration data of the edge cluster of the content delivery network, and perform configuration optimization based on the target static configuration data. The target static configuration data is determined by adopting the target SVM algorithm model with the target outage rate as the target. The target SVM algorithm model is obtained by updating the pre-trained SVM algorithm model based on the target data obtained by static configuration data, dynamic data and parameter tuning step size.

[0086] Specifically, the monitoring platform monitors the outage rate of all domains in the target area in real time and transmits this data to the decision-making platform. The decision-making platform judges the outage rate of the domains at a preset frequency, such as once every 10 minutes. When the outage rate of any domain exceeds the preset outage rate threshold, the network environment of that domain may be abnormal. At this time, the decision-making platform generates an alarm message and sends the alarm message and the target outage rate of the domain to the edge cluster. After the edge cluster determines the target static configuration data corresponding to the target outage rate, it receives the data to optimize the configuration.

[0087] This embodiment provides an SVM-based content delivery network configuration method, which can be used in the central cluster of a content delivery network. Figure 4 This is a flowchart of an SVM-based content delivery network configuration method applied to a central cluster of a content delivery network according to an embodiment of the present invention, such as... Figure 4 As shown, the process includes the following steps:

[0088] Step S401: Monitor the outage rate of all domains in the target region. Specifically, the CDN network covers multiple regions, and the target region is any one of them. The monitoring platform monitors the outage rate of all domains in the target region in real time.

[0089] Step S402: For any domain name, if the domain name's outage rate exceeds a preset outage rate threshold, an alarm message and the target outage rate corresponding to the domain name are sent to the edge cluster of the content delivery network. Specifically, the monitoring platform simultaneously monitors the bandwidth of all domain names. If, with bandwidth remaining constant, the outage rate of any domain name exceeds the preset outage rate threshold, it indicates a network anomaly for that domain name, and an alarm is triggered. The alarm message includes the domain name, time, outage rate, and target region. The monitoring platform transmits this alarm message and the target outage rate for the domain name to the decision platform, which then sends it to the edge cluster corresponding to the target region.

[0090] Step S403: Receive target static configuration data from the edge cluster of the content delivery network. Optimize the configuration based on this target static configuration data. The target static configuration data is determined by using a target SVM algorithm model with a target outage rate as the objective. The target SVM algorithm model is obtained by updating a pre-trained SVM algorithm model based on the target data obtained from the static configuration data, dynamic data, and parameter tuning step size. Specifically, after receiving alarm information and the target outage rate, the edge cluster determines the target static configuration data through steps S201 to S205 and sends it to the decision platform of the central cluster. The central cluster also includes a configuration platform. The decision platform transmits the target static configuration data to the configuration platform, which performs specific configuration operations based on each static configuration parameter and its corresponding data in the target static configuration data. Optionally, the configuration platform can also display the configuration interface during configuration and provide historical configuration records. By configuring according to the target static configuration data, domains with abnormal outage rates reach the target outage rate, restoring the normal network environment and maintaining network stability.

[0091] The content delivery network configuration method based on SVM provided in this embodiment of the invention monitors the interruption rate in real time. When the interruption rate of any domain name exceeds the threshold, an alarm message is promptly sent to the edge cluster. Priority is given to users who discover interruption problems, and target static configuration data sent by the edge cluster is received. Based on this data, configuration optimization is performed to resolve abnormal domain name situations and improve network stability and reliability.

[0092] This embodiment provides an SVM-based content delivery network configuration system, which includes an edge cluster and a central cluster for the content delivery network. Figure 5 This is a flowchart of an SVM-based content delivery network configuration system according to an embodiment of the present invention, such as... Figure 5As shown, the central cluster monitors the outage rate of all domains in the target area. For any domain, if the outage rate exceeds a preset outage rate threshold, the central cluster sends an alarm message and the target outage rate corresponding to the domain to the edge cluster of the content delivery network. The edge cluster receives the alarm message and target outage rate from the central cluster of the content delivery network, obtains the initial static configuration data and dynamic data of all domains in the area where the edge cluster is located, and the alarm message indicates that the domains of the content delivery network are abnormal. The edge cluster determines the parameter tuning step size based on the initial static configuration data of all domains. The edge cluster obtains the target data based on the initial static configuration data, dynamic data, and parameter tuning step size. The edge cluster updates the pre-trained SVM algorithm model based on the target data to obtain the target SVM algorithm model. The edge cluster uses the target SVM algorithm model, with the target outage rate as the target, to determine the target static configuration data, and sends the target static configuration data to the central cluster so that the central cluster can optimize the configuration according to the target static configuration data. The central cluster receives the target static configuration data from the edge cluster of the content delivery network and optimizes the configuration based on the target static configuration data.

[0093] In some optional implementations, the SVM-based content delivery network configuration system includes edge clusters and a central cluster of the content delivery network. The detailed process of this system includes the following steps:

[0094] Step S501: The central cluster monitors the outage rate of all domains in the target region. For details, please refer to [link to relevant documentation]. Figure 4 Step S401 of the illustrated embodiment will not be described again here.

[0095] Step S502: For any domain name, if the domain name's outage rate exceeds a preset outage rate threshold, the central cluster sends an alarm message and the target outage rate corresponding to the domain name to the edge cluster of the content delivery network. For details, please refer to [link to relevant documentation]. Figure 4 Step S402 of the illustrated embodiment will not be described again here.

[0096] Step S503: The edge cluster receives alarm information and the target outage rate from the central cluster of the content delivery network, and obtains the initial static configuration data and dynamic data of all domain names in the region where the edge cluster is located. The alarm information indicates that the domain names of the content delivery network are abnormal. For details, please refer to... Figure 2 Step S201 of the illustrated embodiment will not be described again here.

[0097] Step S504: The edge cluster determines the parameter tuning step size based on the initial static configuration data of all domain names. The initial static configuration data of each domain name includes multiple static configuration parameters and their corresponding data.

[0098] Specifically, step S504 includes:

[0099] Step S5041: Based on the initial static configuration data corresponding to all domain names, determine the average value of each static configuration parameter in the initial static configuration data. Specifically, the static configuration parameters include TCP packet sending strategy, timeout settings, and buffer-related configurations. Taking the timeout setting as an example, assuming there are 3 domain names in the region where the edge cluster is located, obtain the timeout settings for the 3 domain names at the current time: 10 seconds, 15 seconds, and 12 seconds, respectively, and obtain the average timeout setting of 12.33 seconds.

[0100] Step S5042: Determine the standard deviation of the static configuration parameters based on the average value of the static configuration parameters. Specifically, taking the average timeout setting of 12.33 seconds as an example, the standard deviation of the timeout setting is calculated to be 2.055 seconds according to the standard deviation calculation formula.

[0101] Step S5043: Based on the normal distribution and the standard deviation of the static configuration parameters, determine the tuning step size of the static configuration parameters. Specifically, assuming the timeout setting conforms to a normal distribution, the interval between 0 and 1 standard deviation is typically selected as the tuning interval, for example, [12.33-2.055, 12.33+2.055], or [10.275, 14.385]. Divide the tuning interval into several equal parts, and the length of each part is the tuning step size. For example, dividing it into 5 equal parts, with each part having a length of 0.822 seconds, means the tuning step size for the timeout setting is 0.822 seconds. The purpose of using a normal distribution is to quickly obtain the default values ​​of the static configuration data with the highest usage rate of edge nodes in the edge cluster, so as to better suit the current domain name.

[0102] In step S505, the edge cluster obtains target data based on the initial static configuration data, dynamic data, and parameter tuning step size. Specifically, according to the parameter tuning range and step size of the timeout setting, the corresponding value range for the timeout setting is [10.275, 11.097, 11.919, 12.741, 13.563, 14.385]. The TCP packet sending strategy and buffer-related configurations in the initial static configuration data are obtained according to the corresponding parameter tuning step size obtained in step S504 above, resulting in their corresponding value ranges. Combining the initial static configuration data with the dynamic data yields a rich amount of target data for subsequent model updates.

[0103] In step S506, the edge cluster acquires training static data and sample dynamic data, including the outage rate. Specifically, the SVM algorithm model is trained through steps S506 to S510 to obtain a pre-trained SVM algorithm model that accurately predicts the outage rate. More specifically, sample data for training is first acquired, including training static data and sample dynamic data, which are consistent with the data types in step S201.

[0104] In step S507, the edge cluster calculates the average value of the sample dynamic data, excluding the interruption rate, according to time and domain name, to obtain the training dynamic data. Specifically, the sample dynamic data may be data from multiple domain names within a certain historical time period. Therefore, it can be averaged according to time and domain name to obtain the training dynamic data per minute for each domain name, including the average number of TCP connection retransmissions, the average user request header size, the average user response header size, and the average request method.

[0105] Step S508: The edge cluster uses an SVM algorithm model to predict multiple training interruption rates based on training static data and training dynamic data. Specifically, the SVM algorithm model uses the existing SVM (Support Vector Machine) algorithm to predict the training interruption rate corresponding to each set of training static data and training dynamic data. Optionally, the implementation process of the SVM algorithm is existing technology and will not be described in detail here.

[0106] In step S509, the edge cluster determines multiple training errors between multiple training outage rates and outage rates in the training dynamic data. Specifically, each training dynamic data includes the actual outage rate, and the training error is obtained by calculating the variance and standard deviation with the predicted training outage rate.

[0107] In step S510, the edge cluster trains the SVM algorithm model based on multiple training errors to obtain a pre-trained SVM algorithm model. Specifically, using the training error as feedback, an appropriate optimization method is selected to update the model parameters, such as gradient descent or stochastic gradient descent. The adjusted model is then repeated in steps S508 and S509 to ensure that the training error is less than a certain threshold, resulting in the final pre-trained SVM algorithm model. This model can accurately predict the outage rate and has learned the relationship between static configuration data and the outage rate.

[0108] In some alternative implementations, Figure 6 This is a schematic diagram of the SVM algorithm model according to an embodiment of the present invention, as shown below. Figure 6 As shown, static and dynamic training data are input into the SVM algorithm model. The static training data includes TCP packet sending strategy, timeout settings, and buffer-related configurations. The dynamic training data includes average TCP connection retransmissions, average user request header size, average user response header size, average request method, and interruption rate. The SVM algorithm model makes predictions based on the above data to obtain the training interruption rate. The pre-trained SVM algorithm model is then trained based on the input interruption rate and the predicted interruption rate.

[0109] Step S511: The edge cluster updates the pre-trained SVM algorithm model based on the target data to obtain the target SVM algorithm model. The target data includes the interruption rate.

[0110] Specifically, step S511 includes:

[0111] Step S5111: A pre-trained SVM algorithm model is used to predict the target data, resulting in multiple first predicted interruption rates. Specifically, a pre-trained SVM algorithm model is loaded. This model enters a sleep state to conserve CPU resources when no alarm information is received by the edge cluster. Since this model has been trained on sample data, it has learned the relationship between static configuration data and interruption rates. The pre-trained SVM model is used to predict the latest target data, outputting the predicted interruption rate value for each target data point, i.e., the first predicted interruption rate.

[0112] Step S5112: Determine multiple prediction errors between multiple first predicted interruption rates and the interruption rates in the target data. Specifically, the target data includes the actual interruption rates; calculate the variance and standard deviation between the first predicted interruption rates and the actual interruption rates in the target data as prediction errors.

[0113] Step S5113 involves updating the pre-trained SVM algorithm model based on multiple prediction errors to obtain the target SVM algorithm model. Specifically, referring to step S510, the pre-trained SVM algorithm model is updated to obtain the target SVM algorithm model, thereby improving its prediction performance on the new target data and further deepening the understanding of the relationship between static configuration data and the interruption rate.

[0114] In step S512, the edge cluster adopts the target SVM algorithm model, with the target interruption rate as the target, determines the target static configuration data, and sends the target static configuration data to the central cluster so that the central cluster can optimize the configuration according to the target static configuration data.

[0115] In some optional implementations, step S512, the edge cluster employs a target SVM algorithm model, with a target outage rate as the objective, to determine target static configuration data, including:

[0116] Step S5121: Select initial target data from the target data. Specifically, initial target data can be randomly selected from the target data as the starting point for parameter tuning, or the average value of the initial static configuration data in step S5041 and its corresponding dynamic data can be used as the initial target data.

[0117] Step S5122: The target SVM algorithm model is used to predict the initial target data to obtain the second prediction interruption rate. Specifically, the updated target SVM algorithm model is used to predict the initial target data to obtain the second prediction interruption rate.

[0118] In step S5123, if the difference between the second predicted interruption rate and the target interruption rate is greater than a preset error threshold, the static configuration data in the initial target data is adjusted according to the parameter tuning step size. The prediction process is repeated until the difference between the second predicted interruption rate and the target interruption rate is no greater than the preset error threshold. The static configuration data in the target data obtained from the last adjustment is then used as the target static configuration data. Specifically, since the target SVM algorithm model has been updated based on the target data and has relearned the relationship between the static configuration data and the interruption rate, when the target interruption rate is the objective, if the difference between the second predicted interruption rate and the target interruption rate is greater than the preset error threshold, the target SVM algorithm model will automatically perform parameter tuning, that is, change the static configuration data and perform prediction again until the difference between the second predicted interruption rate and the target interruption rate is no greater than the preset error threshold. At this point, the target static configuration data is obtained. This preset error threshold should be as close to 0 as possible so that the second predicted interruption rate is sufficiently close to the target interruption rate. During the automatic parameter tuning process, the target SVM algorithm model increases or decreases the data of each static configuration parameter according to the parameter tuning step size corresponding to each static configuration parameter calculated in step S5043 above.

[0119] Step S513: The central cluster receives the target static configuration data of the edge clusters of the content delivery network and performs configuration optimization based on the target static configuration data. For details, please refer to [link to relevant documentation]. Figure 4 Step S403 of the illustrated embodiment will not be described again here.

[0120] Step S514: The central cluster verifies whether the actual static configuration of the domain name matches the target static configuration data. Specifically, the central cluster also includes a verification platform. The configuration platform adjusts the configuration of domain names with abnormal outage rates based on the target static configuration data received by the decision platform. More specifically, it adjusts the TCP packet sending strategy, timeout settings, and buffer-related configurations according to the target static configuration data to ensure that the outage rate of the domain name reaches the target outage rate and restores the normal network environment. The verification platform is used to monitor the effect of configuration changes and configuration consistency, that is, to verify whether the actual static configuration of the domain name matches the target static configuration data.

[0121] Step S515: If the actual static configuration of the domain name is inconsistent with the target static configuration data, the central cluster automatically repairs the static configuration of the domain name. Specifically, if the actual static configuration is inconsistent with the target static configuration data, the verification platform notifies the configuration platform to perform automatic repair, so that the actual static configuration is consistent with the target static configuration data.

[0122] In some optional implementations, the central cluster also includes a scheduling platform associated with the decision platform. When the decision platform determines that the edge cluster needs to adjust the regional coverage service, it prioritizes adjusting the adjacent nodes in the current region to enter the service and records the change data.

[0123] The content delivery network configuration method based on SVM provided in this invention allows the edge cluster to quickly respond to domain name anomalies by receiving alarm information. It then acquires initial static configuration data and dynamic data, and determines the parameter tuning step size based on the initial static configuration data. This determination of the parameter tuning step size makes the optimization process more refined, flexibly adapting to various situations. Combining the static configuration data, dynamic data, and parameter tuning step size forms target data, providing comprehensive input for subsequent model updates. By updating the SVM algorithm model, it can adapt to new data distributions, improving the model's generalization ability on new data and coping with constantly changing network environments. Since the target SVM algorithm model has learned the relationship between static configuration data and the outage rate, when the target outage rate is the objective, the target SVM algorithm model can accurately determine the target static configuration data through parameter tuning. By receiving alarm information in a timely manner, users are prioritized to discover outage problems, thereby enabling personalized configuration for domain names experiencing anomalies, improving response speed and stability, enhancing the performance of the content delivery network, reducing the outage rate, and improving user experience and satisfaction.

[0124] In this embodiment, the content delivery network configuration device based on SVM is presented in the form of functional units. Here, a unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0125] This invention also provides a computer device; please refer to [link / reference]. Figure 7 , Figure 7 This is a schematic diagram of the structure of a computer device provided in an optional embodiment of the present invention, such as... Figure 7As shown, the computer device includes one or more processors 10, memory 20, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The components communicate with each other via different buses and can be mounted on a common motherboard or otherwise installed as needed. The processors can process instructions executed within the computer device, including instructions stored in or on memory to display graphical information of a GUI on external input / output devices (such as display devices coupled to the interfaces). In some alternative implementations, multiple processors and / or multiple buses can be used with multiple memories and multiple memory modules, if desired. Similarly, multiple computer devices can be connected, each providing some of the necessary operations (e.g., as a server array, a group of blade servers, or a multiprocessor system). Figure 7 Take a processor 10 as an example.

[0126] Processor 10 may be a central processing unit, a network processor, or a combination thereof. Processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The programmable logic device may be a complex programmable logic device (CAMP), a field-programmable gate array (FPGA), a general-purpose array logic (GDA), or any combination thereof.

[0127] The memory 20 stores instructions executable by at least one processor 10 to cause at least one processor 10 to perform the method shown in the above embodiments.

[0128] The memory 20 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the computer device. Furthermore, the memory 20 may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some alternative embodiments, the memory 20 may optionally include memory remotely located relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0129] The memory 20 may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as flash memory, hard disk or solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0130] The computer device also includes a communication interface 30 for communicating with other devices or communication networks.

[0131] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code, which, when accessed and executed by the computer, processor, or hardware, implements the methods shown in the above embodiments.

[0132] A portion of this invention can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to the invention through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0133] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A content delivery network configuration method based on SVM, characterized in that, The method, applied to edge clusters of a content delivery network, includes: The system receives alarm information and target outage rate sent by the central cluster of the content delivery network, and obtains the initial static configuration data and dynamic data of all domain names in the region where the edge cluster is located. The alarm information indicates that the domain name of the content delivery network is abnormal. The initial static configuration data includes the domain name's TCP packet sending strategy, timeout settings and buffer-related configuration. The dynamic data includes the number of retransmissions, outage rate, request method, request header size and response header size of each user request. Based on the initial static configuration data of all domains, determine the parameter tuning step size; Based on the initial static configuration data, the dynamic data, and the parameter tuning step size, the target data is obtained; The pre-trained SVM algorithm model is updated based on the target data to obtain the target SVM algorithm model; Using the target SVM algorithm model and with the target interruption rate as the objective, target static configuration data is determined, and the target static configuration data is sent to the central cluster so that the central cluster can perform configuration optimization according to the target static configuration data.

2. The method according to claim 1, characterized in that, The initial static configuration data for each domain name includes multiple static configuration parameters and their corresponding data; The determination of the parameter tuning step size based on the initial static configuration data of all domain names includes: Based on the initial static configuration data corresponding to all domain names, for each static configuration parameter in the initial static configuration data, the average value of the static configuration parameter is determined; The standard deviation of the static configuration parameters is determined based on the average value of the static configuration parameters. Based on the normal distribution and the standard deviation of the static configuration parameters, the tuning step size of the static configuration parameters is determined.

3. The method according to claim 1, characterized in that, Before updating the pre-trained SVM algorithm model based on the target data to obtain the target SVM algorithm model, the method further includes: Acquire training static data and sample dynamic data, wherein the sample dynamic data includes the interruption rate; The average value of the data in the sample dynamic data, excluding the interruption rate, is calculated according to time and domain name to obtain the training dynamic data; Using the SVM algorithm model, predictions are made based on the static training data and the dynamic training data to obtain multiple training interruption rates; Determine multiple training errors between the multiple training interruption rates and the interruption rates in the training dynamic data; The SVM algorithm model is trained based on the multiple training errors to obtain the pre-trained SVM algorithm model.

4. The method according to claim 1, characterized in that, The target data includes the interruption rate; The process of updating the pre-trained SVM algorithm model based on the target data to obtain the target SVM algorithm model includes: The pre-trained SVM algorithm model is used to predict the target data to obtain multiple first prediction interruption rates; Determine multiple prediction errors between the plurality of first predicted interruption rates and the interruption rates in the target data; The pre-trained SVM algorithm model is updated based on the multiple prediction errors to obtain the target SVM algorithm model.

5. The method according to claim 1, characterized in that, The step of using the target SVM algorithm model, with the target interruption rate as the objective, to determine the target static configuration data includes: Select initial target data from the target data; Using the target SVM algorithm model, the initial target data is predicted to obtain the second prediction interruption rate; If the difference between the second predicted interruption rate and the target interruption rate is greater than a preset error threshold, the static configuration data in the initial target data is adjusted according to the parameter tuning step size, and the prediction process is repeated until the difference between the second predicted interruption rate and the target interruption rate is not greater than the preset error threshold. The static configuration data in the target data obtained from the last adjustment is then used as the target static configuration data.

6. A content delivery network configuration method based on SVM, characterized in that, The method, applied to a central cluster of a content delivery network, includes: Monitor the outage rate of all domains in the target region; For any domain name, if the interruption rate of the domain name exceeds a preset interruption rate threshold, an alarm message and the target interruption rate corresponding to the domain name are sent to the edge cluster of the content delivery network. The system receives target static configuration data for the edge cluster of the content delivery network and optimizes the configuration based on the target static configuration data. The target static configuration data is determined by using a target SVM algorithm model with the target interruption rate as the objective. The target SVM algorithm model is obtained by updating a pre-trained SVM algorithm model based on target data obtained from initial static configuration data, dynamic data, and parameter tuning step size. The initial static configuration data includes the TCP packet sending strategy, timeout settings, and buffer-related configurations for the domain name. The dynamic data includes the retransmission count, interruption rate, request method, request header size, and response header size for each user request.

7. The method according to claim 6, characterized in that, The method further includes: Verify whether the actual static configuration of the domain name is consistent with the target static configuration data; If the actual static configuration of the domain name is inconsistent with the target static configuration data, the static configuration of the domain name will be automatically repaired.

8. A content delivery network configuration system based on SVM, characterized in that, The system includes: Monitor the outage rate of all domains in the target region; For any domain name, if the interruption rate of the domain name exceeds a preset interruption rate threshold, an alarm message and the target interruption rate corresponding to the domain name are sent to the edge cluster of the content delivery network. The system receives alarm information and target outage rate sent by the central cluster of the content delivery network, and obtains the initial static configuration data and dynamic data of all domain names in the region where the edge cluster is located. The alarm information indicates that the domain name of the content delivery network is abnormal. The initial static configuration data includes the domain name's TCP packet sending strategy, timeout settings and buffer-related configuration. The dynamic data includes the number of retransmissions, outage rate, request method, request header size and response header size of each user request. Based on the initial static configuration data of all domains, determine the parameter tuning step size; Based on the initial static configuration data, the dynamic data, and the parameter tuning step size, the target data is obtained; The pre-trained SVM algorithm model is updated based on the target data to obtain the target SVM algorithm model; Using the target SVM algorithm model and with the target interruption rate as the objective, target static configuration data is determined, and the target static configuration data is sent to the central cluster so that the central cluster can perform configuration optimization according to the target static configuration data. Receive the target static configuration data of the edge cluster of the content delivery network, and optimize the configuration based on the target static configuration data.

9. A content delivery network configuration device based on SVM, characterized in that, The device is used in an edge cluster of a content delivery network, and includes: The receiving module is used to receive alarm information and target outage rate sent by the central cluster of the content delivery network, wherein the alarm information indicates that the domain name of the content delivery network is abnormal; The acquisition module is used to acquire the initial static configuration data and dynamic data of all domain names in the region where the edge cluster is located. The initial static configuration data includes the domain name's TCP packet sending strategy, timeout settings, and buffer-related configurations. The dynamic data includes the number of retransmissions, interruption rate, request method, and the request header size and response header size of each user request. The algorithm module is used to determine the parameter tuning step size based on the initial static configuration data of all domain names, obtain target data based on the initial static configuration data, the dynamic data and the parameter tuning step size, update the pre-trained SVM algorithm model based on the target data to obtain the target SVM algorithm model, and use the target SVM algorithm model to determine the target static configuration data with the target interruption rate as the objective. The upload module is used to send the target static configuration data to the central cluster so that the central cluster can perform configuration optimization according to the target static configuration data.

10. A content delivery network configuration device based on SVM, characterized in that, The device, used in a central cluster of a content delivery network, includes: A monitoring platform used to monitor the outage rate of all domains in the target area; The decision platform is used to send alarm information and the target interruption rate corresponding to the domain name to the edge cluster of the content delivery network when the interruption rate of the domain name exceeds a preset interruption rate threshold. It also receives target static configuration data from the edge cluster of the content delivery network and performs configuration optimization based on the target static configuration data. The target static configuration data is determined by using a target SVM algorithm model with the target interruption rate as the objective. The target SVM algorithm model is obtained by updating a pre-trained SVM algorithm model based on target data obtained from initial static configuration data, dynamic data, and parameter tuning step size. The initial static configuration data includes the domain name's TCP packet sending strategy, timeout settings, and buffer-related configurations. The dynamic data includes the number of retransmissions, interruption rate, request method, and the request header size and response header size for each user request.

Images