Content distribution network node scheduling method, electronic device, medium, and program product

Abstract

The application discloses a content distribution network node scheduling method, an electronic device, a medium and a program product, relates to the technical field of network communication, and the content distribution network node scheduling method comprises the following steps: when user request information is received, determining a corresponding node set in a preset three-dimensional mapping diagram according to position information of the user request information, wherein the three-dimensional mapping diagram comprises a corresponding relationship between position information and performance scores of nodes in a content distribution network; inputting a hash value of the user request information and the node set into a preset node scheduling model, and outputting a target node corresponding to the user request information by the node scheduling model; and responding to the user request information through the target node. The technical scheme of the application determines the target node by using the geographical position information and the real-time performance status of the node, effectively improves the efficiency and accuracy of node selection, and reduces the delay of user request and access.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of network communication, and particularly relates to a content distribution network node scheduling method, an electronic device, a computer readable storage medium and a computer program product. BACKGROUND

[0002] The content distribution network is a layer of intelligent virtual network constructed on the basis of the existing Internet by placing node servers everywhere in the network, which can redirect the user's request to the nearest service node according to the network traffic and the connection, load condition, distance to the user and response time of each node and other comprehensive information in real time.

[0003] The current node scheduling strategy of the content distribution network is mainly realized by the resource caching and redistribution, which can better handle static resources, but has poor applicability for real-time data streams, resulting in too high delay of node scheduling. Moreover, the traditional node scheduling strategy is mainly based on proximity and load size, and has limited response capability to the complex real-time conditions of the network, which will affect the access speed, real-time performance and stability of the user.

[0004] The above description is only used to assist understanding of the technical scheme of the present application, and does not represent the acknowledgement of the above description as prior art. SUMMARY

[0005] The main purpose of the present application is to provide a content distribution network node scheduling method, an electronic device, a computer readable storage medium and a computer program product, which aims to solve the technical problem of high delay of the current node scheduling strategy of the content distribution network for user request and access.

[0006] To achieve the above purpose, the present application provides a content distribution network node scheduling method, which comprises:

[0007] When receiving user request information, the corresponding node set is determined in a preset three-dimensional mapping diagram according to the position information of the user request information, wherein the three-dimensional mapping diagram comprises the corresponding relationship between the position information and performance score of each node in the content distribution network;

[0008] The hash value of the user request information and the node set are input into a preset node scheduling model, and the target node corresponding to the user request information is predicted and output by the node scheduling model, wherein the node scheduling model is used to find the best node in the node set according to the hash value;

[0009] The target node responds to the user request information.

[0010] In an embodiment, the step of determining a corresponding node set in the preset three-dimensional mapping according to the location information of the user request information comprises:

[0011] Mapping the location information of the user request information into the three-dimensional mapping to obtain a center point;

[0012] Finding each node in the three-dimensional mapping which has a distance less than a preset distance threshold to the center point to obtain a node set corresponding to each node.

[0013] In an embodiment, the node scheduling model comprises at least an embedding layer, a graph neural network, a graph attention network and a fully connected layer;

[0014] The step of predicting and outputting the target node corresponding to the user request information by the node scheduling model comprises:

[0015] Mapping the hash value to a high-dimensional feature space by the embedding layer to obtain a first feature vector, and determining a feature vector of each node based on the first feature vector, wherein the feature vector of the node comprises at least a distance between the location of the user request information and the performance data of the node itself;

[0016] Performing graph convolution operation on the feature vectors corresponding to each node in the node set by the graph neural network to obtain a second feature vector;

[0017] Inputting the second feature vector into the graph attention network, performing weighted aggregation on the second feature vector by the graph attention network to obtain a third feature vector;

[0018] Inputting the third feature vector into the fully connected layer, combining the performance score and Manhattan distance of each node by the fully connected layer to predict the distribution probability of each node, and selecting the node with the highest probability as the target node.

[0019] In an embodiment, after the step of determining a corresponding node set in the preset three-dimensional mapping according to the location information of the user request information, the method further comprises:

[0020] Calculating a hash value corresponding to the user request information;

[0021] If there is a matching node in the preset hash mapping table which is consistent with the hash value and located in the node set, determining the matching node as the target node corresponding to the user request information, wherein the hash mapping table is used to store the mapping relationship between the hash value of historical user request information and each node;

[0022] Responding to the user request information according to the target node.

[0023] In an embodiment, before the step of receiving the user request information, the method further comprises:

[0024] Collecting original log data of each node in the content distribution network;

[0025] Extracting performance data of multiple dimensions in the original log data, and calculating performance scores of each node according to the performance data corresponding to each node respectively;

[0026] Extracting historical user request information corresponding to each node in the original log data, and converting each historical user request information into a corresponding hash value;

[0027] According to the location information and performance scores of each node, a three-dimensional mapping diagram corresponding to each node is established, and a hash mapping table between each hash value and each node is established.

[0028] In an embodiment, the performance data at least includes one of the capacity, geographical location, margin, temperature, response time, throughput, load capacity and request success rate of the node;

[0029] The step of determining the performance score of each node according to the performance data corresponding to each node respectively comprises:

[0030] Converting each performance data of each node pair into a corresponding feature vector, and performing standardization processing to obtain each standard performance data;

[0031] Calculating the covariance between each standard performance data, which is used to represent the correlation between each standard performance data;

[0032] Eigenvalue decomposition is performed on the covariance matrix to obtain the eigenvalue and eigenvector corresponding to each standard performance data respectively;

[0033] According to the component size of each eigenvector, the weight corresponding to each performance data is determined;

[0034] According to the value and weight of each performance data corresponding to each node, a weighted average is performed to obtain the performance score corresponding to each node.

[0035] In an embodiment, the content distribution network node scheduling method further comprises:

[0036] In response to a configuration instruction, a content distribution network configuration interface is displayed, wherein the content distribution network configuration interface is used to adjust the weight corresponding to each performance data, the distance threshold corresponding to the node set, or the weight ratio of distance to performance for finding the best node according to the received parameter adjustment instruction;

[0037] According to the location information, performance score and load condition of each node in the content distribution network, a network topology map is generated and displayed, wherein different performance scores and load conditions are displayed by visual elements of different colors or sizes.

[0038] In addition, to achieve the above-mentioned purpose, the present application also provides an electronic device, which comprises a memory, a processor and a computer program stored in the memory and executable on the processor, and the computer program is configured to implement the steps of the content distribution network node scheduling method as described above.

[0039] In addition, to achieve the above-mentioned purpose, the present application also provides a storage medium, which is a computer readable storage medium, and a computer program is stored on the storage medium, and the computer program is executed by a processor to implement the steps of the content distribution network node scheduling method as described above.

[0040] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which comprises a computer program, and the computer program is executed by a processor to implement the steps of the content distribution network node scheduling method as described above.

[0041] The present application provides a content distribution network node scheduling method, when receiving a user request information, determining a corresponding node set in a preset three-dimensional mapping according to the location information of the user request information, wherein the three-dimensional mapping includes the corresponding relationship between the location information and the performance score of each node in the content distribution network, then inputting the hash value of the user request information and the node set into a preset node scheduling model, and outputting the target node corresponding to the user request information by the node scheduling model, wherein the node scheduling model is used to find the best node in the node set according to the input hash value, and finally responding to the user request information through the target node. In the technical scheme of the present application, when facing the user request information issued by the user, the corresponding node set is selected based on the corresponding geographical position, and then the best node is found in the node set based on the pre-trained node scheduling model, which combines the geographical position and performance score of the node to determine the best node in the content distribution network, so as to quickly determine the target node to respond to the user request information. Compared with the traditional node scheduling strategy, the real-time dynamic node condition and intelligent algorithm are combined to select the target node, which improves the efficiency and accuracy of node scheduling, and has strong practicability and expansibility. BRIEF DESCRIPTION OF DRAWINGS

[0042] The accompanying drawings, which are incorporated into and form part of the specification, illustrate embodiments consistent with the present application and, together with the specification, serve to explain the principles of the application.

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the accompanying drawings needed to be used in the embodiments or prior art description will be briefly introduced as follows. Obviously, for those skilled in the art, the other drawings can be obtained based on these drawings without any creative effort.

[0044] Figure 1 The flowchart provided by the content distribution network node scheduling method embodiment one of the present application;

[0045] Figure 2 The three-dimensional mapping diagram for representing the mapping relationship between the CDN node position and the performance score in the embodiment of the present application;

[0046] Figure 3 The principle diagram for determining the node set in the three-dimensional mapping diagram in the embodiment of the present application;

[0047] Figure 4 The structure diagram of a feasible node scheduling model in the embodiment of the present application;

[0048] Figure 5 The device structure diagram of the hardware running environment involved in the content distribution network node scheduling method embodiment of the present application.

[0049] The object implementation, functional features and advantages of the present application will be further explained with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0050] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application, and are not used to limit the present application.

[0051] In order to better understand the technical solutions of the present application, the following will be described in detail with reference to the drawings and specific embodiments in the specification.

[0052] The execution subject of the present embodiment can be a computing service device with data processing, network communication and program running functions, such as tablet computer, personal computer, mobile phone, server, etc., or an electronic device, control device, etc. capable of realizing the above functions. The following takes the server as the execution subject as an example to describe the present embodiment and each of the following embodiments.

[0053] The current content distribution strategy is mainly realized through resource caching redistribution. Although it can better handle static resources, it is less applicable to real-time data streams and has high latency. Traditional load balancing strategies mainly schedule nodes based on proximity and load size, but have limited response capability to real-time network conditions, which leads to the fact that traditional CDNs (Content Delivery Network) cannot flexibly respond to network burst conditions, easily leading to uneven resource allocation, affecting user access speed and stability. In addition, traditional CDN technology usually lacks personalized configuration and intelligent scheduling functions, and cannot adaptively dynamically adjust according to user needs and network conditions, which leads to the fact that traditional CDNs have limitations in meeting different user needs and handling complex network conditions, and cannot realize personalized customization and optimization of CDN services.

[0054] In order to overcome the technical problems and defects existing in the prior art, the embodiments of the present application provide a content distribution network node scheduling method, which refers to Figure 1 , Figure 1 The flowchart of the first embodiment of the content distribution network node scheduling method of the present application is shown in the figure. The content distribution network node scheduling method comprises the following steps:

[0055] Step S10, when receiving user request information, determining the corresponding node set in the preset three-dimensional mapping according to the position information of the user request information, wherein the three-dimensional mapping includes the corresponding relationship between the position information and the performance score of each node in the content distribution network.

[0056] The embodiments of the present application can be applied to the management system of the content distribution network (CDN). When the CDN receives new user request information, the location of the user account corresponding to the user request information can be determined first, and the position information is determined. Then, according to the position information of the user, the node set within a certain range is found in the three-dimensional mapping, wherein the positions of the nodes in the node set are relatively close to the position of the user (for example, less than a preset threshold). Because the closer the distance, the faster the response and interaction between the node and the user, the lower the delay, the preliminary screening of each node in the CDN through the position information corresponding to the user request information is realized.

[0057] It should be noted that the three-dimensional mapping is generated by the position information of each node in the content distribution network and the performance score of the real-time performance index. For example, Figure 2 The CDN nodes are represented in the form of a three-dimensional graph "(x, y, z)", wherein the horizontal coordinate x and the vertical coordinate y represent the longitude coordinate and the latitude coordinate of the node position information, and the height z represents the performance score of the node. Such representation can intuitively show the geographical position distribution and performance score of the node, providing data basis for subsequent node scheduling.

[0058] In step S20, the hash value of the user request information and the node set are input into a preset node scheduling model, and a target node corresponding to the user request information is predicted and output by the node scheduling model, wherein the node scheduling model is used to find the best node in the node set according to the hash value;

[0059] In step S30, the target node responds to the user request information.

[0060] In the embodiments of the present application, the node scheduling model (a neural network model) is used to complete the task of finding the best routing node in the node set according to the hash value of the received user request information and in combination with the geographic location and performance score of the node, so as to improve the efficiency and accuracy of node finding.

[0061] It should be noted that the node scheduling model is a neural network model trained in advance by a large amount of training data composed of hash values of user request information and node sets, which can find a target node with a relatively best comprehensive index corresponding to the location and performance data in the node set according to the input hash value, so that the target node scheduled in the CDN can quickly respond after the user sends the user request information, the node selection efficiency and response speed are improved, and the delay is reduced.

[0062] Specifically, the node scheduling model can comprehensively evaluate the performance data and distance (distance between the user's location and the node's location) of each node in the node set according to the input hash value (including location information), and finally output a distribution probability of each node. The higher the probability, the better the node is selected as the target node. When performing comprehensive evaluation, the performance data and distance correspond to a weight ratio respectively. For example, the sum of the weight ratios of the two is 1, and the greater the weight ratio, the more attention is paid to the data in this dimension (performance data or distance). The weight ratio can be iteratively optimized in the model training of node scheduling, or can be customized and configured by the user.

[0063] In addition, the user request information is usually a demand instruction of the user to the CDN to request certain data or media information. After the CDN receives the user request information, the target node is called to respond to the request, and the user required data or information is sent to the user equipment corresponding to the user request information.

[0064] The technical scheme of the embodiments of the present application comprehensively considers the node performance, location distance and user demand in the node selection process by introducing the hash value and node mapping, realizes more intelligent node selection, and improves the performance and user experience of the system.

[0065] The embodiment of the application provides a content distribution network node scheduling method. When receiving user request information, a corresponding node set is determined in a preset three-dimensional mapping diagram according to position information of the user request information, wherein the three-dimensional mapping diagram comprises a corresponding relationship between position information and performance scores of nodes in a content distribution network, then a hash value of the user request information and the node set are input into a preset node scheduling model, a target node corresponding to the user request information is output by the node scheduling model, wherein the node scheduling model is used to find the best node in the node set according to the input hash value, and finally the target node is used to respond to the user request information. In the technical scheme of the embodiment of the application, when facing the user request information, the corresponding node set is selected based on the corresponding geographical position, then the best node is found in the node set based on the pre-trained node scheduling model, the best node in the content distribution network is determined by combining the geographical position and the performance score of the node, so that the target node is quickly determined to respond to the user request information. Compared with the traditional node scheduling strategy, the real-time dynamic node condition and the intelligent algorithm are combined to select the target node, the efficiency and the accuracy of the node scheduling are improved, and the technical scheme has strong practicability and expansibility.

[0066] Further, in a feasible implementation, the step of determining the corresponding node set in the preset three-dimensional mapping diagram according to the position information of the user request information comprises:

[0067] In step S11, the position information of the user request information is mapped into the three-dimensional mapping diagram to obtain a center point.

[0068] In step S12, each node whose distance to the center point is less than a preset distance threshold is found in the three-dimensional mapping diagram to obtain a node set corresponding to each node.

[0069] The embodiment of the application provides a method for screening nodes to determine a node set according to a preset three-dimensional mapping diagram and position information of user request information.

[0070] First, refer to Figure 3 In the process of finding the nearest node set based on the geographical position, a plane formed by an x-axis (longitude coordinate) and a y-axis (latitude coordinate) in the three-dimensional mapping diagram represents the position information, the coordinates of the user position (i.e. the center point) of the user request information after being mapped into the three-dimensional mapping diagram are (x0, y0), the position information of node i is (xi, yi), i is any number in 1 to m, and m is the total number of nodes in the three-dimensional mapping diagram. In the process of determining the node set, the distance D between each node and the user position can be calculated, and the calculation expression is ; wherein, ​is the distance between node i and the user location. Further, a preset distance threshold is R, and a set of all nodes i (including nodes WithinR) satisfying may be obtained, that is, a set N of all nodes within the range R of the user request.

[0071] In a feasible implementation, referring to Figure 4 , the node scheduling model at least includes an embedding layer, a graph neural network, a graph attention network, and a fully connected layer.

[0072] The step of predicting a target node corresponding to the user request information by the node scheduling model includes:

[0073] Step S21, mapping the hash value to a high-dimensional feature space by an embedding layer to obtain a first feature vector, and determining a feature vector of each node based on the first feature vector, wherein the feature vector of the node at least includes the distance between the user request information and the location and the performance data of the node itself.

[0074] The hash value corresponding to the user request is processed and mapped to a high-dimensional feature space. This step is realized by an embedding layer, which converts discrete hash values into continuous vector representations. Assuming that the input hash value is h, the first feature vector obtained after the embedding layer is . The high-dimensional vector can capture the potential relationship between hash values, which helps subsequent feature extraction and matching.

[0075] In addition, the first feature vector at least includes the coordinates of the location of the user request information, and based on the determined positions of the nodes in the node set, the distance between each node in the node set and the location of the user request can be calculated. The performance data of the node can be obtained by collecting real-time data of the CDN network, and the performance data at least includes the capacity, geographic location, margin, temperature, response time, throughput, load capacity, and request success rate of the node.

[0076] Step S22, performing graph convolution operation on the feature vectors corresponding to the nodes in the node set by a graph neural network to obtain a second feature vector.

[0077] In the embodiments of the present application, a graph neural network (GNN) is also introduced to process the topological structure relationship between nodes. Specifically, the node set N can be represented as a graph, where the nodes represent the nodes in the set, and the edges represent the connection relationship between the nodes. The core of the graph neural network is to aggregate the information of the node neighborhood through a graph convolution operation, so as to extract the feature vector of the node. Among them, the feature vector of node i in the graph is , the neighbor node set is N(i), and the new feature vector of node i obtained after the graph convolution operation (i.e. the second feature vector) can be represented as:

[0078] ;

[0079] Among them, is an activation function, W is a weight matrix, and b is a bias vector. It should be noted that the feature vector of node i

[0080] Step S23, inputting the second feature vector into the graph attention network, and performing weighted aggregation on the second feature vector through the graph attention network to obtain a third feature vector;

[0081] In order to further enhance the expression ability of the model, the embodiments of the present application introduce a graph attention network (GAT). The GAT performs weighted aggregation by assigning different attention weights to the neighbor nodes, so that the model can pay attention to the neighbor nodes that are more important to node i. Among them, the attention weight between node i and neighbor node j is , then the new feature vector of node i (i.e. the third feature vector) can be represented as:

[0082] ;

[0083] Among them, the attention weight is calculated from the feature vectors of node i and node j, and can be represented as:

[0084] ;

[0085] Among them, a is a weight vector in the attention mechanism, and || represents the connection operation of the vector.

[0086] Step S24, inputting the third feature vector into the full connection layer, combining the performance score and Manhattan distance of each node through the full connection layer to predict the distribution probability of each node, and selecting the node with the highest probability as the target node.

[0087] In the embodiments of the present application, the features can be mapped to N nodes through the output layer of the neural network, the probability distribution of each node is output, and the node with the highest probability is selected as the best routing node. Specifically, this step is implemented through a fully connected layer (Fully Connected Layer). Assuming that the feature vector of the node is (equal to the third feature vector processed in the foregoing embodiments ), the calculation of the output layer can be represented as:

[0088] ;

[0089] wherein, is a preset weight matrix of the fully connected layer, is a bias vector, and p is a probability distribution vector of the node (i.e., distribution probability). Wherein, characterizes the performance score of the node itself and the Manhattan distance between the location corresponding to the user request information, and through the calculation and processing of the fully connected layer, the distribution probability of each node i can be calculated. The higher the distribution probability, the better the comprehensive evaluation of the performance score and the Manhattan distance, and the more suitable the node is used as the target node of the user request information.

[0090] Further, the method for calculating the weighted Manhattan distance includes that assuming that the geographic location of the user request information is , the geographic location of the node i is , and the self-defined weights are α and β respectively, the distance between the user and the node i can be represented as the sum of the weighted Manhattan distances , and the expression is: .

[0091] In another feasible embodiment, the weighted Manhattan distances of all candidate nodes in the node set are calculated, and then the node with the smallest weighted Manhattan distance is selected as the best node. If the node is inconsistent with the original guide node, dynamic redirection is performed.

[0092] In the embodiments of the present application, the hash value is mapped to a high-dimensional feature space through an embedding layer, the node features are extracted by using a graph neural network and a graph attention network, finally, the probability distribution of the node is output through a fully connected layer, the best node is selected and dynamic redirection is performed. The method of combining the weighted Manhattan distance to select the best node is also disclosed. The scheme of the embodiments of the present application fully utilizes the topological structure and geographic location information between nodes, improves the efficiency and accuracy of node selection, and has strong practicability and expansibility.

[0093] In another feasible embodiment, after the step of determining the corresponding node set in the preset three-dimensional mapping diagram according to the location information of the user request information, the method further includes:

[0094] Step S40, calculating the hash value corresponding to the user request information;

[0095] Step S50, if the hash value is consistent with the hash value of the matching node in the preset hash mapping table and located in the node set, determining the matching node as the target node corresponding to the user request information, wherein the hash mapping table is used to store the mapping relationship between the hash value of the historical user request information and each node;

[0096] Step S60, responding to the user request information according to the target node.

[0097] The embodiment of the present application also provides another way to find the best node in the node set, and steps S40 to S60 are in parallel with steps S20 to S30. After step S10, step S40 can be directly executed, and if the condition described in step S50 is met, the matching node is taken as the target node to realize the fast matching of the target node; if the condition described in step S50 is not met, steps S20 and S30 are executed.

[0098] The hash mapping table is established in advance by the hash values of the user request information corresponding to each node collected, the key of the hash mapping table is the node ID, and the value is the hash value of the corresponding user request information, wherein each node ID can correspond to one or more hash values of the user request information.

[0099] Exemplarily, when receiving a new user request information, if the hash value obtained by calculating the user request information is consistent with the hash value of a historical user request information in the node set, the CDN node routed to by the historical user request information is used as the target node, so that the routing node of most request information can remain unchanged when the node is added or reduced, and the node response efficiency is effectively improved.

[0100] In a feasible embodiment, before the step of receiving the user request information, the method further comprises:

[0101] Step A10, collecting original log data of each node in the content distribution network;

[0102] In the embodiment of the present application, a method for collecting real-time node original data in the content distribution network before processing the newly received user request information is also provided.

[0103] Exemplarily, in the embodiments of the present application, Apache Kafka (a distributed stream processing platform), Apache Flink (a distributed stream processing framework), Influxdb (a time series database), etc. can be used to build an efficient real-time CDN data stream processing and analysis system, so as to realize real-time collection, processing and analysis of CDN log data, and provide reliable data support for subsequent steps.

[0104] Specifically, first, a data collection agent is deployed on each CDN node, and real-time data is collected through the probes and log collectors of each CDN node to capture key network data and user request information, including but not limited to user request data (IP, URL, device information, etc.) and node performance indicators (node response time, throughput, load capacity, request success rate, etc.). Then, Kafka is used as a message queue system to realize real-time high-throughput data collection and transmission. Kafka provides distributed, persistent and highly reliable features, which can meet the real-time processing needs of large-scale CDN data. Kafka producers continuously send these log data to the Kafka cluster. After the data stream reaches Kafka, Apache Flink is used for real-time calculation and processing. Flink provides stream processing capabilities, which can process unlimited data streams and has the characteristics of low latency and high throughput. In Flink, stream processing programs are written, including data cleaning, real-time calculation, aggregation statistics, etc. to extract valuable information and insights from raw log data, and to provide support for subsequent analysis and prediction. The processed result data is stored in the time series database InfluxDB. InfluxDB is a database specially used for storing time series data, which has high performance, scalability and flexible query language, and is very suitable for storing CDN real-time performance data and statistical information for subsequent query and analysis.

[0105] In the embodiments of the present application, stream engine technology is introduced, which enables the content distribution system to process a large number of dispersed data streams in real time, including user requests and node performance information. This is the key to efficient distribution of dynamic content and real-time data streams, and significantly improves the system's ability to process dynamic and real-time content.

[0106] Step A20, extracting performance data of multiple dimensions in the original log data, and calculating performance scores of each node according to the performance data corresponding to each node respectively;

[0107] In the embodiments of the present application, each node in the CDN needs to be scored for performance, which facilitates providing data basis for subsequent node and traffic scheduling processes.

[0108] Specifically, the multi-dimensional performance data of each node is first extracted, including but not limited to the capacity, geographical location, margin, temperature, response time, throughput, load capacity, request success rate and other indicators of the node. These data will serve as the basis for evaluating the performance of the node, helping us to fully understand the running situation and status of each node. Among them, the capacity refers to the size of the storage or processing capacity of the node, usually expressed in terms of storage space or processing capacity; the geographical location refers to the geographical location information of the node, including longitude and latitude; the margin is used to represent the remaining amount of storage or processing capacity of the node, which is used to evaluate the current availability of the node; the temperature refers to the temperature condition of the node, which may affect the performance and stability of the node; the response time refers to the time for the node to respond to a request, usually the time interval from sending a request to receiving a response; the throughput refers to the speed or capacity of the node to handle requests, usually expressed in terms of the number of requests handled per unit of time; the load capacity refers to the current load condition of the node, which is used to evaluate whether the processing capacity of the node has reached the limit; the request success rate refers to the ratio of successful requests handled by the node, which is used to evaluate the stability and reliability of the node.

[0109] In addition, in order to calculate the comprehensive score of each node based on the performance indicator data of each dimension, the embodiments of the present application use the principal component analysis (PCA) method to further process the extracted node performance data to determine the weight corresponding to each dimension performance indicator. Principal component analysis is a commonly used multivariate statistical method that can convert multiple highly correlated indicators into a set of independent principal components, thereby reducing the data dimension and retaining most of the information. Through principal component analysis, it can be determined which performance indicators have a relatively greater impact on the node performance score, and then the weight of each indicator can be determined.

[0110] In another possible embodiment, in addition to using the PCA method, other methods can also be used to determine the weight of the node performance indicators, such as manual setting based on experience or other statistical methods.

[0111] Step A30, extracting historical user request information corresponding to each node in the original log data, and converting each historical user request information into a corresponding hash value;

[0112] In the embodiments of the present application, based on the collected CDN log information, multi-dimensional information about user requests can be parsed, including but not limited to: the geographic location of the user can be obtained by IP address resolution; the device type of the user, using the useragents library according to userAgent including device type (such as desktop, mobile phone, tablet), operating system, browser type and version, etc. Through the analysis result, the device type of the user can be determined; the content type of the request, static resources such as pictures, CSS and JavaScript files, usually have specific file extensions (such as.jpg,.css,.js). Dynamic pages usually contain query parameters in the URL path. Video streams may have specific paths (such as / video / ) or specific extensions (such as.mp4,.m3u8). The Accept and ContentType fields in the HTTP header (Headers) can provide information about the content type that the request expects to receive and the actual content type of the response. For example: Accept: text / html usually indicates that the request is for dynamic content or HTML page; Accept: video / * indicates that the client requests video content; ContentType: application / json indicates that the response content is in JSON format, usually used for API response; using MIME type, MIME type (specified in HTTP through ContentType header field) is an important standard for identifying content type. By configuring the CDN node to parse the response ContentType, the content type can be accurately identified, including static resources, dynamic content and video streams, etc.

[0113] Further, after extracting the historical request information of each dimension, the multi-dimensional request information of the user is converted into a corresponding hash value using a hash algorithm. For example, a common hash algorithm SHA-1 can be used to convert the string form of user information into a fixed length hash value. The calculation formula of the hash algorithm is as follows:

[0114] In this step, we use a hash algorithm to convert the multi-dimensional information of the user into a hash value. This hash value will represent the user's request information for subsequent node matching and selection. We use a common hash algorithm such as SHA-1 to convert the string form of user information into a fixed length hash value, which ensures that each user request information will generate a hash value, ensuring the recognizability and consistency of the request.

[0115] The calculation formula of the hash algorithm is as follows: wherein, UserInfo represents the multi-dimensional information of the user, and H() represents the hash algorithm.

[0116] Step A40, according to the position information and performance score of each node, a three-dimensional mapping diagram corresponding to each node is established, and a hash mapping table between each hash value and each node is established.

[0117] In the process of establishing the three-dimensional mapping diagram, the CDN nodes can be represented in the form of a three-dimensional diagram "(x, y, z)", wherein the abscissa x and the ordinate y represent the longitude and latitude of the node, and the height z represents the performance score of the node in step two. The three-dimensional mapping diagram can intuitively show the geographical position distribution and performance score of the node, and provide a data basis for subsequent node scheduling.

[0118] In the process of establishing the hash mapping table, all nodes in the CDN need to be traversed, and the hash value of the user multi-dimensional information calculated in step A30 is mapped with the node ID to construct a hash mapping table. The key of the hash mapping table is the node ID, and the value is the hash value of the corresponding user request information. The hash mapping table can be used for subsequent traffic scheduling and node selection. The correspondence between the key and the value in the hash mapping table is shown in the following formula: , wherein, represents the ID of node i.

[0119] Further, in a feasible embodiment, the performance data at least includes one of capacity, geographical position, margin, temperature, response time, throughput, load capacity and request success rate of the node;

[0120] The step of determining the performance score of each node according to the performance data corresponding to each node includes:

[0121] Step A21, converting each performance data of each node pair into a corresponding feature vector, and performing standardization processing to obtain each standard performance data;

[0122] Step A22, calculating the covariance between each standard performance data, which is used to represent the correlation between each standard performance data;

[0123] Step A23, performing eigenvalue decomposition on the covariance matrix to obtain the eigenvalue and eigenvector corresponding to each standard performance data;

[0124] Step A24, determining the weight corresponding to each performance data according to the component size of each eigenvector;

[0125] Step A25, performing weighted average according to the value and weight of each performance data corresponding to each node to obtain the performance score corresponding to each node.

[0126] The embodiment of the application provides a method for calculating the performance score of each node by using a principal component analysis method. Specifically, the performance indicators of the nodes are defined as m-dimensional vectors where i represents the ith node, and m represents the number of performance indicators. For example, the performance indicators can include the capacity of the node, geographical location, response time, throughput, load capacity, request success rate, etc.

[0127] Further, the node performance data needs to be standardized to ensure that each performance indicator has the same quantity, for example, the standardization formula is as follows:

[0128] ;

[0129] wherein, is the standardized node performance data, is the original data, is the mean of the jth indicator, is the standard deviation of the jth indicator.

[0130] The covariance between the standardized node performance data is further calculated to obtain the correlation between the node performance indicators. The formula for calculating the covariance matrix is:

[0131] ;

[0132] wherein Cov(X) is the covariance matrix, n is the sample size, and Z is the standardized node performance data matrix.

[0133] The next step is to calculate the eigenvalues and eigenvectors. By performing eigenvalue decomposition on the covariance matrix, the eigenvalues λ1, λ2, …, λm and the corresponding eigenvectors v1, v2, …, vm can be obtained, wherein the eigenvector vi represents the weight vector associated with the ith principal component. Finally, according to the size of the components of the eigenvectors, the weights of each performance indicator are determined. It should be noted that the weights of each performance indicator can also be customized in the individual configuration platform.

[0134] After determining the weights of each performance indicator, a weighted method can be used to score the performance of each node and perform normalization to ensure that the scores are on a unified scale. The weighted method can weight the sum of the performance data of the node according to the weights of each indicator to obtain the comprehensive node performance score. Normalization can map the score to a unified interval, such as 0 to 100, to facilitate subsequent comparison and analysis.

[0135] wherein the weighted average is to multiply the value of each performance indicator by its corresponding weight, and then sum all the weighted values. Specifically, for the ith node, the comprehensive performance score is calculated according to the following formula:

[0136] ;

[0137] wherein, represents the weight of the jth performance indicator, represents the value of the jth performance indicator of the ith node, and m represents the number of performance indicators. It should be noted that when the value of the performance indicator is inversely proportional to the comprehensive performance score, for example, the longer the node response time, the lower the performance score, the performance score cannot be directly calculated by the length of the node response time, and needs to be pre-processed (such as taking the reciprocal or taking the opposite number) first, so that its monotonicity is consistent with the monotonicity of the performance score.

[0138] In addition, in order to ensure that the scores are on a unified scale, the obtained comprehensive performance scores are subjected to linear normalization processing to map the scores to a specified interval, for example, 0 to 100, and the calculation formula is as follows:

[0139] ;

[0140] wherein, MinScore and MaxScore are the minimum and maximum values of all node scores, and MinRange and MaxRange are the minimum and maximum values of the target interval after normalization.

[0141] Through the above weighted average and normalization processing, the comprehensive performance score of each node can be obtained, which ensures that the scores of the nodes are on a unified scale, facilitates subsequent comparison and analysis, and thus selects the best node.

[0142] The embodiments of the present application introduce a principal component analysis method to process node performance data to determine the weight corresponding to each performance indicator. Through PCA, a plurality of node performance indicators with high correlation can be converted into a group of principal components that are independent of each other, thereby reducing the data dimension and retaining most of the information. This processing method helps to comprehensively understand the running situation and state of the node, and provides important data support for intelligent scheduling and load balancing.

[0143] In another possible embodiment, the content distribution network node scheduling method further comprises:

[0144] Step B10, in response to the configuration instruction, a content distribution network configuration interface is displayed, wherein the content distribution network configuration interface is used to adjust the weight corresponding to each performance data, the distance threshold corresponding to the node set, or the weight ratio of distance and performance for finding the best node according to the received parameter adjustment instruction;

[0145] Step B20, generating and displaying a network topology map according to the position information, performance score and load situation of each node in the content distribution network, wherein different performance scores and load situations are displayed with different colors or sizes of visual elements.

[0146] The embodiments of the present application also provide a method for using a personalized configuration tool to make the parameter configuration of a content distribution network system more flexible. Illustratively, a content distribution network configuration interface can be provided to users in the form of a personalized configuration tool, aiming to allow users to personalize and adjust various parameters of the CDN system, and can also include weights corresponding to various node performance indicators, pre-defined rules in abnormal situations, location range R, weight ratio of distance and performance, etc., while providing a visual presentation of the network topology and node performance of the CDN to help achieve intelligent scheduling and traffic balancing optimization of the CDN nodes.

[0147] Illustratively, the front-end interface can be built first, and a user-friendly single-page application (SPA) can be built using Vue.js to provide an intuitive configuration interface. Through a graphical method, users can quickly customize and adjust the CDN service policy. The front-end interface is simple and clear, and has good user interaction experience.

[0148] In addition, in order to show the geographical location and related information of the nodes, the embodiments of the present application can adopt the form of a map, which can be realized on the front-end interface by integrating a map component library such as Google Maps API or Mapbox. Specifically, the geographical location of each CDN node is marked on the map, and the performance score and load condition of each node are added. Such a map display method can allow users to intuitively understand the distribution and performance of the nodes, which helps them make more intelligent configuration and adjustment decisions.

[0149] Further, on the basis of the map display, the performance score and load condition of the nodes can also be distinguished by visual elements such as color and size. For example, different colors can be used to represent the performance level of the nodes, or different sizes of markers can be used to represent the load condition of the nodes. Such visual presentation method can allow users to more intuitively understand the overall situation of the nodes and quickly make corresponding operations and adjustments.

[0150] In the front-end part, the Axios library (a network request library) is integrated to handle HTTP requests and realize asynchronous data exchange with the back-end. The front-end sends a configuration request to the back-end and receives the response data processed by the back-end, realizing real-time data interaction between the front-end and the back-end.

[0151] The backend part uses a Spring Boot framework (an open-source application framework) to develop RESTful API (application programming interface based on the REST (Representational State Transfer) architectural style) services to handle configuration requests initiated by the front end. The backend uses the powerful processing capability of Spring Boot to parse and verify the configuration data submitted by the user and convert it into policy rules that can be understood by the CDN scheduling system.

[0152] In terms of data storage and management, the use of a relational database such as MySQL (relational database management system) for configuration data storage is considered to ensure persistent storage and efficient management of configurations. Spring Data JPA is used to simplify database interactions, enabling quick querying and updating of configuration data.

[0153] The application embodiments also provide a real-time feedback and application mechanism to ensure that configuration modifications take effect immediately. WebSocket communication is implemented through the Java backend to provide real-time feedback to users about the effects of configuration changes. Users can immediately see the impact of configuration modifications, improving the accuracy and efficiency of configurations.

[0154] Access control and permission management are implemented through the security framework provided by Spring Security to ensure the security of configuration tools. Only authorized users can access and modify configuration data, ensuring the security and stability of the system.

[0155] The application embodiments provide an efficient, secure, and easy-to-use CDN service policy configuration platform for users by introducing a personalized configuration tool. It supports quick customization and adjustment of CDN service policies to meet the performance optimization needs of different users. It also allows users to customize CDN service policies according to their own needs, achieving personalized customization and optimization of CDN services, improving the flexibility and adaptability of the system, and meeting the needs of users for efficient, intelligent, and personalized content distribution.

[0156] The technical solutions of the above embodiments not only solve the deficiencies of existing CDN technology in dynamic content processing, intelligent traffic management, and user service experience, but also promote the development of CDN technology towards more efficient, more intelligent, and more user-friendly directions.

[0157] It should be noted that the above examples are for understanding the application and do not limit the content distribution network node scheduling method of the application. Further simple transformations based on this technical concept are within the scope of protection of the application.

[0158] The application provides an electronic device, comprising: at least one processor; and a memory connected with the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the content distribution network node scheduling method in the above embodiment one.

[0159] Reference is made below in conjunction with Figure 5 which shows a structural schematic diagram of an electronic device suitable for implementing the embodiments of the application. The electronic device in the embodiments of the application can include, but is not limited to, mobile terminals such as mobile phones, notebook computers, digital broadcast receivers, PDAs (Personal Digital Assistant), PADs (Portable Application Description), PMPs (Portable Media Player), vehicle-mounted terminals (for example, vehicle-mounted navigation terminals), and the like, and fixed terminals such as digital TVs, desktop computers, and the like. Figure 5 The electronic device shown is merely an example and should not bring any limitation to the functions and use range of the embodiments of the application.

[0160] As Figure 5 shown, the electronic device can include a processing device 1001 (for example, a central processor, a graphics processor, and the like) which can perform various appropriate actions and processes according to programs stored in a read-only memory (ROM) 1002 or loaded from a storage device 1003 into a random access memory (RAM) 1004. In the RAM 1004, various programs and data required for the operation of the electronic device are also stored. The processing device 1001, the ROM 1002, and the RAM 1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: input devices 1007 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, and the like; output devices 1008 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, and the like; the storage device 1003 including, for example, a magnetic tape, a hard disk, and the like; and a communication device 1009. The communication device 1009 can allow the electronic device to communicate with other devices wirelessly or by wire to exchange data. Although the electronic device with various systems is shown in the figure, it should be understood that all the systems shown are not required to be implemented or possessed. More or less systems can be alternatively implemented or possessed.

[0161] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product comprising a computer program carried on a computer readable medium, the computer program containing program codes for executing the method shown in the flowchart. In such embodiments, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiments disclosed in the present application are executed.

[0162] The electronic device provided by the present application adopts the content distribution network node scheduling method in the above-mentioned embodiments, which can solve the technical problem of high delay of the current content distribution network node scheduling strategy for user requests and access. Compared with the prior art, the electronic device provided by the present application has the same beneficial effects as the content distribution network node scheduling method provided by the above-mentioned embodiments, and other technical features in the electronic device are the same as the features disclosed in the previous embodiment method, which will not be repeated here.

[0163] It should be understood that parts of the present application can be realized by hardware, software, firmware or a combination thereof. In the description of the above-mentioned embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0164] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0165] The present application provides a computer readable storage medium having stored thereon computer readable program instructions (i.e. computer program) for executing the content distribution network node scheduling method in the above-mentioned embodiments.

[0166] The computer readable storage medium provided in the present application may, for example, be a U disk, but is not limited to an electric, magnetic, optical, electromagnetic, infrared, or semiconductor system, system, or device, or any combination of the above. More specific examples of the computer readable storage medium can include, but are not limited to, an electric connection with one or more conductive wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present embodiment, the computer readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer readable storage medium can be transmitted by any suitable medium, including but not limited to an electric wire, an optical cable, an RF (Radio Frequency), and the like, or any suitable combination of the above.

[0167] The above computer readable storage medium can be included in an electronic device, or can exist separately without being assembled into an electronic device.

[0168] The above computer readable storage medium carries one or more programs, which, when executed by an electronic device, cause the electronic device to: when receiving user request information, determine a corresponding node set in a preset three-dimensional mapping according to position information of the user request information, wherein the three-dimensional mapping includes a corresponding relationship between position information and performance scores of each node in a content distribution network; input a hash value of the user request information and the node set into a preset node scheduling model, and output a target node corresponding to the user request information by the node scheduling model, wherein the node scheduling model is used to find the best node in the node set according to the hash value; and respond to the user request information through the target node.

[0169] Computer program code for carrying out operations of the present application can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C++ or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider).

[0170] The flow diagrams and the block diagrams in the drawings are illustrations of architectures, functionalities, and operations of possible implementations of systems, methods, and computer program products according to various embodiments of the present application. In this regard, each block in the flow diagrams or block diagrams can represent a module, a segment, or a portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that in some alternative implementations, the functions noted in the blocks can occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently or the blocks may

[0171] The modules involved in the embodiments of the present application can be implemented in the form of software or in the form of hardware. In some cases, the name of the module does not constitute a limitation on the module itself.

[0172] The readable storage medium provided by the present application is a computer readable storage medium, which stores computer readable program instructions (i.e., a computer program) for executing the content distribution network node scheduling method described above, and can solve the technical problem of high delay of the current content distribution network node scheduling strategy for user requests and access. Compared with the prior art, the computer readable storage medium provided by the present application has the same beneficial effects as the content distribution network node scheduling method provided by the above-mentioned embodiments, and will not be described here.

[0173] The application also provides a computer program product comprising a computer program which, when executed by a processor, implements the steps of the content distribution network node scheduling method as described above.

[0174] The computer program product provided by the application can solve the technical problem of high delay of the current content distribution network node scheduling strategy for user requests and access. Compared with the prior art, the beneficial effects of the computer program product provided by the application are the same as those of the content distribution network node scheduling method provided by the above-mentioned embodiments, and are not described here.

[0175] The above only describes some embodiments of the application, and does not limit the patent scope of the application. Any equivalent structural transformation, direct / indirect application in other related technical fields based on the technical concept of the application, and the contents of the specification and drawings are included in the patent protection scope of the application.

Claims

1. A content distribution network node scheduling method, characterized by, The content distribution network node scheduling method comprises: When receiving user request information, determining a corresponding node set in a preset three-dimensional mapping diagram according to position information of the user request information, wherein the three-dimensional mapping diagram comprises a corresponding relationship between position information and performance scores of nodes in a content distribution network; inputting a hash value of the user request information and the node set into a preset node scheduling model, and outputting a target node corresponding to the user request information by the node scheduling model, wherein the node scheduling model is used to find the best node in the node set according to the hash value; responding to the user request information through the target node; The node scheduling model at least comprises an embedding layer, a graph neural network, a graph attention network and a fully connected layer; The step of outputting the target node corresponding to the user request information by the node scheduling model comprises: mapping the hash value to a high-dimensional feature space through the embedding layer to obtain a first feature vector, and determining a feature vector of each node based on the first feature vector, wherein the feature vector of the node at least comprises a distance between the position of the user request information and its own performance data; performing graph convolution operation on the feature vectors corresponding to the nodes in the node set through the graph neural network to obtain a second feature vector; inputting the second feature vector into the graph attention network, and performing weighted aggregation on the second feature vector through the graph attention network to obtain a third feature vector; inputting the third feature vector into the fully connected layer, combining the performance scores and Manhattan distances of the nodes through the fully connected layer to predict the distribution probability of each node, and selecting the node with the highest probability as the target node.

2. The content distribution network node scheduling method of claim 1, wherein, The step of determining the corresponding node set in the preset three-dimensional mapping diagram according to the position information of the user request information comprises: mapping the position information of the user request information to the three-dimensional mapping diagram to obtain a center point; finding nodes in the three-dimensional mapping diagram with a distance less than a preset distance threshold from the center point to obtain a node set corresponding to the nodes.

3. The content distribution network node scheduling method of claim 1, wherein, After the step of determining the corresponding node set in the preset three-dimensional mapping diagram according to the position information of the user request information, the method further comprises: calculating a hash value corresponding to the user request information; if a matching node in the node set exists in a preset hash mapping table and is consistent with the hash value, determining the matching node as the target node corresponding to the user request information, wherein the hash mapping table is used to store a mapping relationship between hash values of historical user request information and nodes; responding to the user request information according to the target node.

4. The content distribution network node scheduling method of claim 1, wherein, Before the step of receiving user request information, the method further comprises: collecting original log data of nodes in a content distribution network; extracting performance data in multiple dimensions from the original log data, and calculating performance scores of the nodes according to the performance data corresponding to each node respectively; extracting historical user request information corresponding to each node in the original log data, and converting each historical user request information into a corresponding hash value; establishing a three-dimensional mapping diagram corresponding to each node according to the position information and performance score of each node, and establishing a hash mapping table between each hash value and each node.

5. The content distribution network node scheduling method of claim 4, wherein, The performance data at least includes one of capacity, geographical position, margin, temperature, response time, throughput, load capacity and request success rate of the node; The step of determining the performance score of each node according to the performance data corresponding to each node comprises: Converting each performance data of each node pair into a corresponding feature vector and performing standardization processing to obtain each standard performance data; Calculating the covariance between each standard performance data, which is used to represent the correlation between each standard performance data; Performing eigenvalue decomposition on the covariance matrix to obtain the eigenvalue and eigenvector corresponding to each standard performance data; Determining the weight corresponding to each performance data according to the component size of each eigenvector; Performing weighted average on the value and weight of each performance data corresponding to each node to obtain the performance score corresponding to each node.

6. The content distribution network node scheduling method of any one of claims 1 to 5, wherein, The content distribution network node scheduling method further comprises: In response to a configuration instruction, a content distribution network configuration interface is displayed, wherein the content distribution network configuration interface is used to adjust the weight corresponding to each performance data, the distance threshold corresponding to the node set, or the weight ratio of distance to performance for finding the best node according to the received parameter adjustment instruction; According to the position information, performance score and load condition of each node in the content distribution network, a network topology map is generated and displayed, wherein different performance scores and load conditions are displayed with different colors or sizes of visual elements.

7. An electronic device, comprising: The device comprises a memory, a processor and a computer program stored on the memory and executable on the processor, the computer program being configured to implement the steps of the content distribution network node scheduling method according to any one of claims 1 to 6.

8. A storage medium, characterized by The storage medium is a computer readable storage medium, and the storage medium stores a computer program, which is executed by the processor to implement the steps of the content distribution network node scheduling method according to any one of claims 1 to 6.

9. A computer program product, characterised in that, The computer program product comprises a computer program, which is executed by the processor to implement the steps of the content distribution network node scheduling method according to any one of claims 1 to 6.

Images