Scheduling method, device and electronic equipment of content distribution network

Abstract

The application discloses a scheduling method and device of a content distribution network and electronic equipment, and relates to the technical field of content distribution networks. The method comprises the following steps: acquiring CDN log data in a historical preset time period; analyzing offline data to obtain a relationship table of access duration and access quality scores; processing performance data to obtain a node stability classifier; acquiring a predicted access duration of user access to a target node; based on the predicted access duration, the relationship table of access duration and access quality scores, and the node stability classifier, calculating the stability feature and the access quality score of the target node; and correcting a scheduling strategy according to the stability feature and the access quality score of the target node. The historical CDN log data is used as a data basis, the actual situation of user access to a node is considered, and from the two aspects of access quality and stability, the reliability of future access to each node is predicted and the conventional scheduling strategy is corrected, so that a node with better service quality is selected.

Description

Technical Field

[0001] This application relates to the field of content delivery network technology, and specifically to a scheduling method, apparatus, and electronic device for a content delivery network. Background Technology

[0002] CDN (Content Delivery Network) scheduling refers to directing all existing network requests for a client's domain name to a suitable target data center, thereby achieving traffic control, quality control, cost control, and fault handling. Commonly used CDN scheduling strategies include DNS (Domain Name System) scheduling, HTTP scheduling, and bandwidth- and load-based scheduling methods.

[0003] Among these methods, scheduling based on DNS and HTTP protocols follows the proximity principle, selecting the nearest node server based on geographical location. However, proximity alone cannot guarantee optimal service quality for users. Scheduling based on bandwidth and load only detects snapshots of the current network conditions. In unstable environments, snapshots cannot reflect the true network status, lacking consideration of actual circumstances. Therefore, current CDN scheduling methods suffer from a lack of consideration for the actual conditions of nodes, failing to guarantee optimal service quality for each node. Summary of the Invention

[0004] In view of this, this application provides a scheduling method, apparatus and electronic device for a content delivery network, the main purpose of which is to solve the problem that the current CDN scheduling method lacks consideration of the actual situation of the nodes and cannot guarantee the best quality of service for the nodes.

[0005] Firstly, this application provides a scheduling method for a content delivery network, comprising:

[0006] Acquire CDN log data within a preset historical time period; the CDN log data includes offline data and performance data.

[0007] The offline data was analyzed to obtain a table showing the relationship between access duration and access quality score;

[0008] The performance data is processed to obtain a node stability classifier; the node stability classifier is used to determine the stability characteristics of a node based on its current load.

[0009] Obtain the predicted access duration for users to access the target node;

[0010] Based on the predicted access duration, the relationship table between the access duration and the access quality score, and the node stability classifier, the stability characteristics and access quality score of the target node are calculated; wherein, the number of the target nodes is at least one.

[0011] The scheduling strategy is modified based on the stability characteristics and access quality scores of the target nodes. Optionally, modifying the scheduling strategy based on the stability characteristics and access quality scores of the target nodes includes: obtaining the stability characteristics and access quality scores of each node accessed by the user based on the predicted access duration; wherein the stability characteristics include high and low; sorting the target nodes in descending order according to the access quality scores, and removing target nodes with low stability characteristics to obtain a node access priority list; and modifying the scheduling strategy based on the node access priority list.

[0012] Optionally, based on the predicted access duration, the stability characteristics of user access to each node are obtained, including: using the predicted access duration to calculate the workload of each node within the corresponding time period; using the workload of each node as the input of the node stability classifier to obtain the stability characteristics of each node.

[0013] Optionally, the step of analyzing the offline data to obtain a table showing the relationship between access duration and access quality score includes: aggregating preset fields of the offline data to obtain score evaluation indicators for accessing each node; wherein the preset fields include source address, destination address, access duration, and access time period; the score evaluation indicators include hit rate, response time, and HTTP status code; and performing weighted calculation based on the score evaluation indicators to obtain a table showing the relationship between access duration and access quality score.

[0014] Optionally, processing the performance data to obtain a node stability classifier includes: filtering the performance data to obtain metrics for measuring node load; the metrics include at least one of memory, traffic, bandwidth, number of requests, number of access failures, and duration; aggregating the metrics by node address and using them as input to a pre-trained node stability prediction model to obtain the node stability classifier.

[0015] Optionally, the step of filtering the performance data to obtain metrics for measuring node load includes: cleaning the performance data as input to a big data computing engine to obtain the metrics from the performance data that can be used to measure node load.

[0016] Optionally, obtaining the predicted access duration of a user's access to a target node includes: processing CDN log data within the historical preset time period using a time prediction function to obtain the processing results of the access time and access duration of each target node in history; and predicting the access duration of the user's access to the target node based on the processing results to obtain the predicted access duration.

[0017] Secondly, this application provides a scheduling device for a content delivery network, comprising:

[0018] The acquisition unit is configured to acquire CDN log data within a preset historical time period; the CDN log data includes offline data and performance data.

[0019] The analysis unit is configured to analyze the offline data to obtain a table showing the relationship between access duration and access quality score;

[0020] The processing unit is configured to process the performance data to obtain a node stability classifier; the node stability classifier is used to determine the stability characteristics of the node based on the current load of the server node.

[0021] The prediction unit is configured to obtain the predicted access duration of the user to the target node;

[0022] The calculation unit is configured to calculate the stability features and access quality score of the target node based on the predicted access duration, the relationship table between the access duration and the access quality score, and the node stability classifier; wherein the number of the target nodes is at least one.

[0023] The correction unit is configured to correct the scheduling strategy based on the stability characteristics and access quality score of the target node.

[0024] Thirdly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the content delivery network scheduling method described in the first aspect.

[0025] Fourthly, this application provides an electronic device, including a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, wherein the processor executes the computer program to implement the content delivery network scheduling method described in the first aspect.

[0026] Using the above technical solution, this application provides a content delivery network scheduling method, apparatus, and electronic device. First, CDN log data within a historical preset time period is acquired. The CDN log data includes offline data and performance data. The offline data is analyzed to obtain a relationship table between access duration and access quality score. The performance data is processed to obtain a node stability classifier. The node stability classifier is used to determine the stability characteristics of a node based on its current load. The predicted access duration of a user accessing a target node is obtained, and the stability characteristics and access quality score of the target node are calculated based on the access duration, the relationship table between access duration and access quality score, and the node stability classifier. The number of target nodes is at least one. The scheduling strategy is modified based on the stability characteristics and access quality score of the target nodes. This application uses historical CDN log data as its primary data foundation to analyze user access behavior and habits, thereby better reflecting the actual situation of the nodes accessed by users. It also comprehensively predicts the network reliability of each node accessed by users in the future from the perspectives of access quality and stability. Based on the given network reliability, that is, to evaluate from the perspectives of access quality score and stability, the conventional scheduling strategy is modified to select nodes with better service quality.

[0027] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0028] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0029] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0030] Figure 1 A flowchart illustrating a scheduling method for a content delivery network provided in an embodiment of this application is shown.

[0031] Figure 2 A schematic diagram of scheduling based on the HTTP protocol provided in an embodiment of this application is shown;

[0032] Figure 3 A schematic diagram of load-based scheduling provided in an embodiment of this application is shown;

[0033] Figure 4 A flowchart illustrating another content delivery network scheduling method provided in an embodiment of this application is shown.

[0034] Figure 5 This illustration shows a schematic diagram of the structure of a content delivery network scheduling device provided in an embodiment of this application. Detailed Implementation

[0035] To better understand the above-mentioned objectives, features, and advantages of this application, the solution of this application will be further described below. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0036] To address the issue that current CDN scheduling methods lack consideration for the actual conditions of nodes and cannot guarantee optimal service quality for nodes, this embodiment provides a content delivery network scheduling method, such as... Figure 1 As shown, the method includes:

[0037] S101, retrieve CDN log data within a historical preset time period.

[0038] CDN log data includes offline data and performance data. Offline data includes information such as the IP address of each CDN access; performance data includes information on CPU, memory, and bandwidth usage. More specifically, it refers to CDN server offline logs and CDN server performance logs. The historical preset time period here refers to CDN log data within a certain period of time in the past, such as collecting CDN log data from users over the past 7 days or the past 14 days.

[0039] S102, Analyze the offline data to obtain a table showing the relationship between access duration and access quality score.

[0040] By performing a series of analyses on offline data, a relationship table between access duration and access quality score can be obtained. This table is used to derive the corresponding access quality score based on the node accessed by the user and the duration of access to that node. In other words, given the access duration, the access quality score corresponding to that access duration can be obtained, making it convenient to combine this table with subsequent predicted access durations to obtain the corresponding access quality score.

[0041] S103 processes the performance data to obtain a node stability classifier.

[0042] The performance data is processed to obtain a node stability classifier, which is used to determine the stability characteristics of a node based on its current load. Specifically, in S102 and S103, different processing methods are applied to the two types of CDN data to obtain a relationship table for access quality scores and a node stability classifier.

[0043] S104, obtain the predicted access duration of the user to the target node.

[0044] Obtaining the predicted access duration of a user's access to a target node refers to predicting the access time of a user to a certain target node in the future. The number of target nodes is at least one, meaning that the user may access multiple nodes in the future, and the prediction is made about the access duration of the user's access to each node.

[0045] S105, based on the predicted access duration, the relationship table between access duration and access quality score, and the node stability classifier, calculate the stability features and access quality score of the target node.

[0046] Furthermore, based on the access duration, the stability characteristics and access quality score of the target node are calculated using the relation table and stability classifier obtained in S102 and S103.

[0047] S106. Based on the stability characteristics and access quality score of the target node, the scheduling strategy is modified.

[0048] In this embodiment, CDN log data within a preset historical time period is first acquired. This CDN log data includes offline data and performance data. The offline data is analyzed to obtain a table showing the relationship between access duration and access quality score. The performance data is processed to obtain a node stability classifier. This classifier is used to determine the stability characteristics of a node based on its current load. The predicted access duration for a user accessing a target node is obtained, and the stability characteristics and access quality score of the target node are calculated based on the access duration, the table showing the relationship between access duration and access quality score, and the node stability classifier. There is at least one target node. The scheduling strategy is then adjusted based on the stability characteristics and access quality score of the target node. This embodiment uses historical CDN log data as the primary data foundation to analyze user access behavior and habits, thus better reflecting the actual situation of the nodes accessed by the user. It comprehensively predicts the network reliability of each node accessed by the user in the future from both access quality and stability perspectives. Based on the given network reliability, that is, an evaluation is conducted from both access quality score and stability aspects, and the conventional scheduling strategy is adjusted to select nodes with better service quality.

[0049] Optionally, the scheduling strategy is modified based on the stability characteristics and access quality scores of the target nodes, including: obtaining the stability characteristics and access quality scores of each node accessed by the user based on the predicted access duration; wherein the stability characteristics include high and low; sorting the target nodes in descending order according to the access quality scores, and removing the target nodes with low stability characteristics to obtain a node access priority list; and modifying the scheduling strategy based on the node access priority list.

[0050] In this embodiment, after predicting the user's access duration to the target node, the access quality score corresponding to the access duration at a future time can be retrieved based on the relationship table between access duration and access quality score. Simultaneously, based on this access duration, a node stability classifier is used to obtain the stability characteristics of the access duration to the target node. Then, the nodes are sorted in descending order of their access quality scores, and nodes with low stability are removed from the sorted sequence. Thus, the remaining nodes in the sequence are all stable and usable, and possess both stability and access quality score priority criteria, facilitating the adjustment of scheduling strategies and the selection of suitable service nodes. The overall approach is based primarily on the user's historical CDN log data. Through multi-dimensional aggregation analysis and machine learning training on the log data, the criteria for determining stability and access quality scores are obtained. Furthermore, based on historical log data, the system predicts when a user might access a specific node and for how long in the future. This prediction is then combined with stability and access quality scores to prioritize users' access to each node based on their stability and access quality scores (and unstable nodes are removed). This results in a priority list for each node, which is then used to adjust the conventional scheduling strategy. This addresses the current CDN scheduling methods' lack of consideration for the actual conditions of nodes, thus failing to guarantee optimal service quality for each node.

[0051] Optionally, based on the predicted access duration, the stability characteristics of each node accessed by the user are obtained, including: using the predicted access duration to calculate the workload of each node within the corresponding time period; using the workload of each node as the input of the node stability classifier to obtain the stability characteristics of each node.

[0052] In this embodiment, since the node stability classifier is used to determine the stability characteristics of a node based on its current load, it is necessary to first calculate the workload of accessing the corresponding node in the future time period based on the predicted access duration, and then use the workload of each node as the input to the node stability classifier to obtain the stability characteristics of each node. The purpose of the node stability classifier is to derive a formula relating server load and stability, aiming to calculate the probability of server instability based on the load index. This probability is converted into a stability characteristic, including two judgment results: high and low. That is, P(stable|load).

[0053] Furthermore, according to the Naive Bayes formula, we can obtain:

[0054]

[0055] To give P (stability) a clear definition, records where the total number of failures in one hour exceeds 10% of the total number are marked as having an unstable load status for that IP during that time period, and vice versa.

[0056] To simplify the calculation model, the peak bandwidth obtained in the above steps is compared with the server's physical bandwidth to obtain the peak physical bandwidth utilization rate; traffic and request counts are compared with the maximum value in the past 7 days to obtain the ratio between the current traffic and the maximum value. The ratios of all the above data are divided into 10 intervals, and different characteristics are labeled for the values ​​in different intervals. Each indicator is limited to 10 characteristic values. The results are shown in Table 1:

[0057]

[0058] Table 1: Load Metric Feature Extraction Rules

[0059] Then it can be defined as:

[0060] P(load) = P(CPU, memory, bandwidth, traffic, number of accesses)

[0061] Furthermore, we can obtain:

[0062]

[0063] Assuming the server's physical bandwidth is 800Mbps, the maximum traffic over 7 days is 200GB, and the maximum number of requests is 500, the example obtained from the above steps can be represented by the following input and output:

[0064] 10.110.150.2 - 0:00 to 0:59 - Traffic 35.23G - Peak bandwidth 20.08M / s - Peak CPU utilization 50% - Peak memory utilization 86% - 56 requests - 2 failed.

[0065] Converting to input and output means:

[0066]

[0067] Table 2: Input and Output Examples of Stability Classifier

[0068] Optionally, offline data is analyzed to obtain a table showing the relationship between access duration and access quality score. This includes: aggregating preset fields of offline data to obtain score evaluation indicators for accessing each node; wherein the preset fields include source address, destination address, access duration, and access time period; the score evaluation indicators include hit rate, response time, and HTTP status code; and performing weighted calculations based on the score evaluation indicators to obtain a table showing the relationship between access duration and access quality score.

[0069] In this embodiment, the step of analyzing offline data to obtain a relationship table for access quality scores first involves aggregating preset fields in the offline data. These preset fields include source address, target access node address, access duration, and access time period, which represent the user's access to content within a preset historical time period, thus obtaining the user's access habits. Further aggregation processing yields score evaluation indicators for each access node, including hit rate, response time, and HTTP status code. These score evaluation indicators are then weighted to obtain a relationship table between access duration and access quality score. By extrapolating more realistic user access habits based on historical data, a more accurate scheduling strategy can be provided.

[0070] Optionally, the performance data is processed to obtain a node stability classifier, including: filtering the performance data to obtain metrics for measuring node load; the metrics include at least one of memory, traffic, bandwidth, number of requests, number of access failures, and duration; aggregating the metrics by node address and using them as input to a pre-trained node stability prediction model to obtain a node stability classifier.

[0071] In this embodiment, the process of obtaining the node stability classifier includes filtering performance data to obtain metrics for measuring node load, then aggregating node addresses and using these as input to a pre-trained node stability prediction model to obtain the node stability classifier. It should be noted that since some metrics include content that overlaps with offline data, such as memory, traffic, bandwidth, and request count, in practice, CDN log data can be processed as a whole. Only the aggregation keywords need to be selected; the appropriate keywords are determined based on actual needs to improve computational efficiency.

[0072] Optionally, the performance data can be filtered to obtain metrics for measuring node load, including: cleaning the performance data as input to a big data computing engine to obtain metrics from the performance data that can be used to measure node load.

[0073] In this embodiment, performance data can be filtered by feeding it into a big data computing engine to quickly obtain metrics for measuring node load, thereby improving computational efficiency.

[0074] Optionally, obtaining the predicted access duration of a user's access to a target node includes: processing CDN log data within a historical preset time period using a time prediction function to obtain the processing results of the access time and access duration of each target node in history; and predicting the access duration of a user's access to a target node based on the processing results to obtain the predicted access duration.

[0075] In this embodiment, the CDN log data within a preset historical time period includes the user's access time, access duration, and access content to each server node during that time period. This data is processed, and the access duration of the user's access to the target node is predicted based on the processing results, resulting in a predicted access duration. In other words, by understanding the user's access habits based on historical log data, the user's access behavior (predicted access duration) can be predicted, thereby enabling the deduction of a more realistic and accurate scheduling strategy based on historical data.

[0076] It is worth mentioning that CDN (Content Delivery Network) scheduling is a common scheduling strategy. Among the common solutions, there are DNS scheduling (Domain Name System), HTTP scheduling, and scheduling methods based on bandwidth and load.

[0077] DNS scheduling: Traditional CDN solutions rely on DNS resolution for scheduling, using the LocalDNS exit address as the query IP and scheduling users geographically based on the query IP, adhering to the proximity principle. Methods based on the HTTP protocol, such as... Figure 2 As shown, the scheduling center server obtains the public IP address of the internet user from the client's request, and uses this as the basis for CDN scheduling to select edge cache service nodes, thus guiding the user to the nearest node; its essence is still the proximity principle. Methods based on bandwidth and load are as follows... Figure 3 As shown, the real-time bandwidth information of all service nodes in the network and the real-time health coefficient of the cluster are obtained through the decision center as load data, which is then used to allocate and schedule nodes.

[0078] Therefore, current CDN scheduling strategies based on DNS and HTTP protocols follow the proximity principle, but this principle cannot guarantee optimal service quality for users. Bandwidth and load-based scheduling methods rely on snapshots of the network's current state; however, in unstable environments, snapshots may not reflect the true network condition. All of the above CDN scheduling strategies lack consideration for the actual situation of the nodes accessed by users.

[0079] In this embodiment, the primary data foundation is based on various historical CDN logs of users. This allows for understanding user access habits and predicting user access behavior (predicting access duration), demonstrating a deeper understanding of the actual situation of user access nodes. Furthermore, through analysis and processing of log data (including multi-dimensional aggregation analysis and machine learning training), two dimensions of node judgment rules are obtained: stability and access quality scores. Finally, based on historical log data, predictions are made about when a user might access a specific node in the future, and for how long. This prediction information is combined with stability and access quality scores to obtain a priority ranking of the stability and access quality scores of user access to each node (and unstable nodes are removed), i.e., obtaining a scheduling priority ranking list for each node. Finally, the conventional scheduling strategy is modified based on the scheduling priority list. This solves the problem that current CDN scheduling methods lack consideration of the actual situation of nodes and cannot guarantee optimal service quality for nodes.

[0080] Furthermore, such as Figure 4 The diagram shown illustrates a scheduling method for another content delivery network provided in this application. The method uses a preset historical time period of the past 7 days as an example to illustrate the scheduling method proposed in this embodiment. The method includes:

[0081] S401 collects CDN log data from users over the past 7 days.

[0082] In this step, CDN log data is actually recorded in an aggregate manner. Therefore, by obtaining the user's CDN log data from the past 7 days, the log data from 7 days ago to 7 days later can be compared, and the incremental data is the user's CDN log data from the past 7 days. By analyzing the user's historical data, we can better understand the user's access habits across different nodes.

[0083] S402 aggregates CDN log data according to different keywords to obtain a node stability classifier and a table showing the relationship between access duration and access quality score.

[0084] CDN log data includes offline data and performance data. Since some aggregation results also involve information such as the target node's memory, bandwidth, traffic, and access count, this embodiment performs unified aggregation processing on the CDN log data. However, it should be noted that the results obtained by aggregation processing based on different keywords are different. Therefore, the specific processing procedures for CDN log data are different when obtaining the node stability classifier and the table showing the relationship between access duration and access quality score.

[0085] Specifically, the process of obtaining a node stability classifier involves filtering performance data from CDN logs (using a big data computing engine for cleaning) to obtain metrics for measuring node load. These metrics include at least one of the following: memory, traffic, bandwidth, number of requests, number of access failures, and duration. These metrics are then aggregated and used as input to a pre-trained node stability prediction model to obtain the node stability classifier. The node stability classifier is then used to determine the stability characteristics of a node based on its current load.

[0086] The process of obtaining the relationship table between access duration and access quality score includes: aggregating preset fields of offline data to obtain the score evaluation indicators for each node accessed; the preset fields include source address, destination address, access duration, and access time period; the score evaluation indicators include hit rate, response time, and HTTP status code. The key to the aggregation process in this step is the preset fields, which are then weighted and calculated based on the score evaluation indicators to obtain the relationship table between access duration and access quality score.

[0087] S403 processes CDN log data using a time prediction function to obtain the predicted access duration for users accessing the target node.

[0088] The CDN log data within a preset historical time period includes the user's access time, access duration, and access content to each server node during that time period. This data is processed, and the access duration for the user's target node is predicted based on the processing results. In other words, by understanding the user's access habits based on historical log data, the user's access behavior (predicting access duration) can be predicted.

[0089] S404, combined with the predicted access duration, calculates the stability and access quality score of the user's future access to the target node.

[0090] S405: Sort the target nodes in descending order according to the access quality score to obtain a priority sequence, and remove nodes with low stability characteristics from the priority sequence.

[0091] S406, Modify the scheduling strategy according to the priority sequence.

[0092] Then, based on the conventional scheduling strategy, the scheduling strategy is modified according to the priority sequence. Taking traditional DNS scheduling as an example: after the local DNS sends a resolution request to the scheduling server and obtains the server node round-robin pool, it matches the client_ip and server_ip pool in the priority list. If the number of matched entries reaches a certain proportion of the round-robin pool capacity (for example, if the round-robin pool size is 20, and 10 data entries are required to be matched in the list, then the condition is met), it is considered that the data training result fits the current network status, and the next step is to directly derive the round-robin order based on the score from high to low. Otherwise, it is considered that the training result is not sufficient to support the correction work under the current network snapshot status, and the round-robin priority calculation continues according to the conventional scheduling strategy.

[0093] This embodiment employs a scheduling method based on historical data for correction. Building upon conventional network probing, it uses server load and peer-to-peer historical access data as standards for correction. This retains the real-time advantage of conventional probing while allowing for the derivation of more realistic and accurate scheduling strategies based on historical data. Furthermore, leveraging the richness of the CDN log structure, it achieves quality prediction scoring. By using linear regression training to predict user access behavior (predicting access duration), it elevates the scoring granularity to the level of predicted access duration, providing a more precise basis for adjusting the scheduling strategy.

[0094] Furthermore, as Figures 1 to 4 The specific implementation of the method shown in this embodiment provides a scheduling device for a content delivery network, such as... Figure 5 As shown, the device includes: an acquisition unit 51, an analysis unit 52, a processing unit 53, a prediction unit 54, a calculation unit 55, and a correction unit 56.

[0095] Acquisition unit 51 is configured to acquire CDN log data within a historical preset time period; the CDN log data includes offline data and performance data.

[0096] Analysis unit 52 is configured to analyze the offline data to obtain a table showing the relationship between access duration and access quality score;

[0097] The processing unit 53 is configured to process the performance data to obtain a node stability classifier; the node stability classifier is used to determine the stability characteristics of the node based on the current load of the server node.

[0098] Prediction unit 54 is configured to obtain the predicted access duration of the user's access to the target node;

[0099] The calculation unit 55 is configured to calculate the stability features and access quality score of the target node based on the predicted access duration, the relationship table between the access duration and the access quality score, and the node stability classifier; wherein the number of the target nodes is at least one.

[0100] The correction unit 56 is configured to correct the scheduling strategy based on the stability characteristics and access quality score of the target node.

[0101] In a specific application scenario, the correction unit 56 is specifically configured to obtain the stability characteristics and access quality scores of each node accessed by the user based on the predicted access duration; wherein, the stability characteristics include high and low; sort the target nodes in descending order according to the access quality scores, and remove the target nodes with low stability characteristics to obtain a node access priority list; and correct the scheduling strategy according to the node access priority list.

[0102] In specific application scenarios, the correction unit 56 is further configured to use the predicted access duration to calculate the workload of each node within the corresponding time period; and use the workload of each node as the input of the node stability classifier to obtain the stability features of each node.

[0103] In a specific application scenario, the analysis unit 52 is further configured to aggregate the offline data based on preset fields to obtain the score evaluation index for accessing each node; wherein, the preset fields include source address, destination address, access duration, and access time period; the score evaluation index includes hit rate, response duration, and HTTP status code; and a weighted calculation is performed based on the score evaluation index to obtain a relationship table between the access duration and the access quality score.

[0104] In a specific application scenario, the processing unit 53 is further configured to filter the performance data to obtain metrics for measuring node load; the metrics include at least one of memory, traffic, bandwidth, number of requests, number of access failures, and duration; the metrics are aggregated by node address and used as input to a pre-trained node stability prediction model to obtain the node stability classifier.

[0105] In specific application scenarios, the processing unit 53 is further configured to use the performance data as input to the big data computing engine for cleaning and processing, so as to obtain the indicator items in the performance data that can be used to measure the node load.

[0106] In specific application scenarios, the correction unit 56 is further configured to process the CDN log data within the historical preset time period through a time prediction function to obtain the processing results of the access time and access duration of each target node in history; based on the processing results, the access duration of the user accessing the target node is predicted to obtain the predicted access duration.

[0107] It should be noted that other corresponding descriptions of the functional units involved in the content delivery network scheduling device provided in this embodiment can be found in [reference]. Figures 1 to 4 The corresponding description in [the document] will not be repeated here.

[0108] Based on the above, Figures 1 to 4 Accordingly, this embodiment also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method. Figures 1 to 4 The method shown.

[0109] Based on this understanding, the technical solution of this application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as CD-ROM, USB flash drive, mobile hard drive, etc.) and includes several instructions to cause a computer device (such as personal computer, server, or network device, etc.) to execute the methods of various implementation scenarios of this application.

[0110] Based on the above, Figures 1 to 4 The method shown, and Figure 5 To achieve the above objectives, this application also provides an electronic device, which can be configured on a computer side, etc. The device includes a storage medium and a processor; the storage medium is used to store a computer program; the processor is used to execute the computer program to achieve the above-described objectives. Figures 1 to 4 The method shown.

[0111] Based on the above, Figures 1 to 4 The method shown, and Figure 5 To achieve the above objectives, the present application also provides a chip in the illustrated virtual device embodiment, including one or more interface circuits and one or more processors; the interface circuits are used to receive signals from the memory of an electronic device and send the signals to the processors, the signals including computer instructions stored in the memory; when the processor executes the computer instructions, it causes the electronic device to perform the above-described... Figures 1 to 4 The method shown.

[0112] Optionally, the aforementioned physical devices may also include a user interface, a network interface, a camera, radio frequency (RF) circuitry, sensors, audio circuitry, a Wi-Fi module, etc. The user interface may include a display screen, input units such as a keyboard, etc., and optional user interfaces may also include USB interfaces, card reader interfaces, etc. The network interface may optionally include standard wired interfaces, wireless interfaces (such as Wi-Fi interfaces), etc.

[0113] Those skilled in the art will understand that the physical device structure provided in this embodiment does not constitute a limitation on the physical device, and may include more or fewer components, or combine certain components, or have different component arrangements.

[0114] The storage medium may also include an operating system and a network communication module. The operating system is a program that manages the hardware and software resources of the aforementioned physical device, supporting the operation of information processing programs and other software and / or programs. The network communication module is used to enable communication between the various components within the storage medium, as well as communication with other hardware and software in the information processing physical device.

[0115] Through the above description of the implementation methods, those skilled in the art can clearly understand that this application can be implemented using software plus necessary general-purpose hardware platforms, or it can be implemented using hardware. By applying the scheme of this embodiment, CDN log data within a historical preset time period is obtained; the CDN log data includes offline data and performance data; the offline data is analyzed to obtain a table showing the relationship between access duration and access quality score; the performance data is processed to obtain a node stability classifier; the node stability classifier is used to determine the stability characteristics of a node based on its current load; the predicted access duration of a user accessing a target node is obtained, and based on the access duration, the stability and access quality score of the target node are calculated; the number of target nodes is at least one; the scheduling strategy is modified based on the stability and access quality score of the target node. This embodiment uses historical CDN log data as the main data foundation to analyze user access behavior and habits, thereby better reflecting the consideration of the actual situation of the nodes accessed by the user; and comprehensively predicts the network reliability of each node accessed by the user in the future from the perspectives of access quality and stability. Based on the given network reliability, that is, evaluating from the perspectives of access quality score and stability, the conventional scheduling strategy is modified to select nodes with better service quality.

[0116] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the term "comprising" or any other variations thereof is intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0117] The above description is merely a specific embodiment of this application, enabling those skilled in the art to understand or implement this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments described herein, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A scheduling method for a content delivery network, characterized in that, include: Retrieve CDN log data within a preset historical time period; The CDN log data includes offline data and performance data; The offline data was analyzed to obtain a table showing the relationship between access duration and access quality score; The performance data is processed to obtain a node stability classifier; The node stability classifier is used to determine the stability characteristics of a node based on its current load. Obtain the predicted access duration for users to access the target node; Based on the predicted access duration, the relationship table between the access duration and the access quality score, and the node stability classifier, the stability characteristics and access quality score of the target node are calculated; wherein, the number of the target nodes is at least one. The scheduling strategy is modified based on the stability characteristics and access quality score of the target node.

2. The method according to claim 1, characterized in that, The step of modifying the scheduling strategy based on the stability characteristics and access quality score of the target node includes: Based on the predicted access duration, the stability characteristics and access quality score of the user's access to each node are obtained; wherein, the stability characteristics include high and low. The target nodes are sorted in descending order according to their access quality scores, and target nodes with low stability characteristics are removed to obtain a node access priority list. The scheduling strategy is modified based on the node access priority list.

3. The method according to claim 2, characterized in that, Based on the predicted access duration, the stability characteristics of user access to each node are obtained, including: Using the predicted access duration, calculate the workload of each node within the corresponding time period; The workload of each node is used as the input to the node stability classifier to obtain the stability characteristics of each node.

4. The method according to claim 1, characterized in that, The analysis of the offline data yields a table showing the relationship between access duration and access quality score, including: The offline data is aggregated using preset fields to obtain a score evaluation index for accessing each node; wherein, the preset fields include source address, destination address, access duration, and access time period; and the score evaluation index includes hit rate, response time, and HTTP status code. Based on the weighted calculation of the scoring evaluation indicators, a relationship table between the access duration and the access quality score is obtained.

5. The method according to claim 1, characterized in that, The process of processing the performance data to obtain a node stability classifier includes: The performance data is filtered to obtain metrics for measuring node load; the metrics include at least one of memory, traffic, bandwidth, number of requests, number of access failures, and duration. The index items are aggregated by node address and used as input to a pre-trained node stability prediction model to obtain the node stability classifier.

6. The method according to claim 5, characterized in that, The process of filtering the performance data to obtain metrics for measuring node load includes: The performance data is used as input to a big data computing engine for cleaning and processing to obtain the metrics that can be used to measure node load.

7. The method according to claim 1, characterized in that, Obtain the predicted access duration for the user to the target node, including: The CDN log data within the historical preset time period is processed using a time prediction function to obtain the processing results of the access time and access duration of each target node in history. Based on the processing results, the access duration of the user to the target node is predicted to obtain the predicted access duration.

8. A scheduling device for a content delivery network, characterized in that, include: The acquisition unit is configured to acquire CDN log data within a preset historical time period. The CDN log data includes offline data and performance data; The analysis unit is configured to analyze the offline data to obtain a table showing the relationship between access duration and access quality score; The processing unit is configured to process the performance data to obtain a node stability classifier; The node stability classifier is used to determine the stability characteristics of a node based on its current load. The prediction unit is configured to obtain the predicted access duration of the user to the target node; The calculation unit is configured to calculate the stability features and access quality score of the target node based on the predicted access duration, the relationship table between the access duration and the access quality score, and the node stability classifier; wherein the number of the target nodes is at least one. The correction unit is configured to correct the scheduling strategy based on the stability characteristics and access quality score of the target node.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 7.

10. An electronic device comprising a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 7.

Images