Access request control method and device, computer program product and electronic equipment

Abstract

The invention discloses an access request control method and device, a computer program product and electronic equipment. The method comprises the following steps: acquiring access characteristics of a target service processing access request in a preset period; inputting the access feature into a target model to obtain a predicted access feature of the target service processing access request in a target time period adjacent to the preset period; acquiring an actual access feature of the target service processing access request in the target time period, and calculating a relative difference between the predicted access feature and the actual access feature to obtain a first difference value; and determining an access request control strategy based on a comparison result between the first difference value and a preset difference threshold, and executing the access request control strategy on the target service. According to the invention, the problem of low response efficiency of the current limiting strategy of the service in the related technology is solved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence, and more specifically, to a method, device, computer program product, and electronic device for controlling access requests. Background Art

[0002] The rapid advancement of internet technology and the increasing diversification of the market have driven innovation in service models. Microservices, with its exceptional modularity, have become an indispensable component of modern software development. The microservices design philosophy emphasizes decomposing complex applications into a series of small, focused service components. These components operate independently, communicate, and collaborate through precisely defined interfaces, achieving system-wide flexibility and scalability. In a highly interconnected microservices ecosystem, services rely on each other to achieve a seamless user experience and efficient service delivery. However, the complexity of microservices architectures also presents new challenges. Within this architecture, performance issues or failures in a single service can rapidly ripple through the entire system, causing a chain reaction. Given the dynamic and uncertain nature of internet applications, services can experience overload or unexpected errors without warning, posing a serious threat to system stability and user experience.

[0003] Related technologies for limiting service traffic flow rely on static rules and lack adaptability to real-time workload changes. In the volatile internet environment, fixed thresholds and rules may not be able to respond promptly to sudden increases in traffic or other dynamic changes, resulting in delayed or ineffective governance measures, increasing system risks and operational burdens.

[0004] Currently, no effective solution has been proposed to address the problem of low response efficiency of the current limiting strategy of services in related technologies. Summary of the Invention

[0005] The main purpose of this application is to provide a method, device, computer program product and electronic device for controlling access requests to solve the problem of low response efficiency of the flow limiting strategy of services in related technologies.

[0006] To achieve the above-mentioned objectives, according to one aspect of the present application, a method for controlling access requests is provided. The method comprises: collecting access features of access requests processed by a target service within a preset period, wherein the preset period includes multiple time periods; inputting the access features into a target model to obtain predicted access features of access requests processed by the target service within a target time period adjacent to the preset period, wherein the target model is trained using multiple sets of training samples, each set of training samples including access features within a historical period and access features within a next time period adjacent to the historical period, the predicted access features including at least one of the following: request rate, request response time, and request failure rate; collecting actual access features of access requests processed by the target service within the target time period, calculating the relative difference between the predicted access features and the actual access features to obtain a first difference value; determining an access request control policy based on a comparison result between the first difference value and a preset difference threshold, and executing the access request control policy on the target service.

[0007] Optionally, determining the access request control strategy based on the comparison result between the first difference value and the preset difference threshold includes: for each predicted access feature, when the first difference value is greater than or equal to the first difference threshold, determining closing the access link as the access request control strategy, wherein the access link is the link for the user to access the target service; when the first difference value is less than the first difference threshold and greater than or equal to the second difference threshold, determining controlling the access frequency to be less than the first frequency threshold as the access request control strategy, wherein the first difference threshold is greater than the second difference threshold, and the access frequency is the number of access requests to access the target service within the target time length; when the first difference value is less than the second difference threshold and greater than or equal to the third difference threshold, determining controlling the access frequency to be less than the second frequency threshold as the access request control strategy, wherein the second frequency threshold is greater than the first frequency threshold, and the second difference threshold is greater than the third difference threshold; when the first difference value is less than the third difference value, determining not controlling the access frequency as the access request control strategy.

[0008] Optionally, after executing the access request control policy on the target service, the method further includes: determining the execution time of the access request control policy and the target access characteristics that require the execution of the access request control policy; obtaining the repair strategy associated with the target access characteristics, and issuing a prompt message, wherein the prompt information includes at least the execution time, the target access characteristics and the repair strategy, and the repair strategy is used to repair the target service.

[0009] Optionally, after executing the access request control policy on the target service, the method further includes: monitoring the access characteristics of the target service within a preset time period to obtain a monitored access characteristic; calculating the relative difference between the monitored access characteristic and the predicted access characteristic to obtain a second difference value; determining whether the second difference value is less than or equal to a target difference threshold; if the second difference value is greater than the target difference threshold, continuing to execute the step of executing the access request control policy on the target service; if the second difference value is less than or equal to the target difference threshold, stopping executing the access request control policy.

[0010] Optionally, the target model is trained in the following manner: obtaining historical access records of the target service, extracting access features within multiple historical periods from the historical access records, and extracting access features within the next historical period of each historical period from the historical access records; determining the access features within each historical period and the access features within the next historical period of the historical period as a set of training samples to obtain multiple sets of training samples; training a neural network model based on the multiple sets of training samples to obtain the target model.

[0011] Optionally, the method also includes: re-acquiring multiple sets of updated training samples every target period, training the target model based on the updated sets of training samples, and determining whether the training of the updated target model is completed; if the training of the updated target model is completed, inputting the access features into the updated target model to obtain updated predicted access features; if the training of the updated target model is not completed, executing the step of inputting the access features into the target model to obtain predicted access features for access requests processed by the target service within the target time period.

[0012] Optionally, collecting access features of the target service processing access requests within a preset period includes: obtaining a feature set of the target service processing access requests within the preset period, preprocessing the feature set to obtain multiple target features, wherein the preprocessing includes at least one of the following: removing duplicate data, outlier detection, missing value processing, data normalization and feature engineering; screening baseline features associated with the predicted access features from multiple target features, and for each target feature, calculating the similarity between the target feature and the baseline feature; when the similarity is within a preset similarity range, determining the target feature as an access feature.

[0013] To achieve the above-mentioned purpose, according to another aspect of the present application, a device for controlling access requests is provided. The device includes: a collection unit for collecting access features of access requests processed by a target service within a preset period, wherein the preset period includes multiple time periods; an input unit for inputting the access features into a target model to obtain predicted access features of access requests processed by the target service within a target time period adjacent to the preset period, wherein the target model is trained by multiple sets of training samples, each set of training samples including access features within a historical period and access features within a next time period adjacent to the historical period, and the predicted access features including at least one of the following: request rate, request response time, and request failure rate; a calculation unit for collecting actual access features of access requests processed by the target service within the target time period, calculating the relative difference between the predicted access features and the actual access features to obtain a first difference value; and a first determination unit for determining an access request control policy based on a comparison result between the first difference value and a preset difference threshold, and executing the access request control policy on the target service.

[0014] In order to achieve the above-mentioned purpose, according to another aspect of the present application, a computer program product is provided, including a computer program, which, when executed by a processor, implements the steps of the access request control method described in each embodiment of the present application.

[0015] Through the present application, the following steps are adopted: collecting access features of the target service processing access requests within a preset period, wherein the preset period includes multiple time periods; inputting the access features into the target model to obtain predicted access features of the target service processing access requests within a target time period adjacent to the preset period, wherein the target model is trained by multiple groups of training samples, each group of training samples includes access features within a historical period and access features within a next time period adjacent to the historical period, and the predicted access features include at least one of the following: request rate, request response time, and request failure rate; collecting actual access features of the target service processing access requests within the target time period, calculating the relative difference between the predicted access features and the actual access features, and obtaining a first difference value; determining an access request control strategy based on the comparison result between the first difference value and a preset difference threshold, and executing the access request control strategy on the target service, thereby solving the problem of low response efficiency of the flow limiting strategy of the service in the related art. By deploying the target model and predicting access characteristics in real time, setting thresholds based on the difference between the predicted results and the actual access characteristics, and designing and initiating access request control policies, we ensure that proactive measures can be taken to prevent impending performance degradation or failures. This reduces the need for manual intervention and significantly enhances the microservice architecture's ability to handle emergencies, enabling applications to operate continuously and stably in complex and changing environments. This, in turn, improves the responsiveness of the service's rate-limiting policies. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The accompanying drawings, which constitute part of this application, are intended to provide a further understanding of this application. The exemplary embodiments and descriptions of this application are intended to explain this application and do not constitute an improper limitation on this application. In the accompanying drawings:

[0017] Figure 1 is a flowchart of a method for controlling access requests provided in an embodiment of the present application;

[0018] Figure 2 This is a schematic diagram of a cloud-native service governance architecture provided according to an embodiment of the present application;

[0019] Figure 3 is a schematic diagram of the data collection and storage process provided according to an embodiment of the present application;

[0020] Figure 4 is a schematic diagram of a data preprocessing module provided according to an embodiment of the present application;

[0021] Figure 5 is a schematic diagram of an optional access request control method provided according to an embodiment of the present application;

[0022] Figure 6 is a schematic diagram of a control device for access requests provided according to an embodiment of the present application;

[0023] Figure 7 is a schematic diagram of an electronic device provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0024] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in this application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0025] In order to enable those skilled in the art to better understand the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of this application.

[0026] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application described here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0027] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for display, data for analysis, etc.) involved in this disclosure are all information and data authorized by the user or fully authorized by all parties.

[0028] It should be noted that the information collected is information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of relevant data comply with the relevant laws, regulations and standards of the relevant regions, take necessary confidentiality measures, do not violate public order and good customs, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0029] The present invention will be described below in conjunction with preferred implementation steps. Figure 1 is a flow chart of a method for controlling access requests according to an embodiment of the present application. Figure 1 As shown, the method includes the following steps:

[0030] Step S101 : collecting access characteristics of a target service processing access request within a preset period, wherein the preset period includes multiple time periods.

[0031] In step S101, access features can be collected through Prometheus (an open source monitoring alarm system and time series database). The access request control method of this embodiment is implemented based on the cloud native service governance architecture. Figure 2 This is a schematic diagram of the cloud native service governance architecture provided by the embodiment of the present application, such as Figure 2 As shown in the figure, the architecture includes business applications, Nacos (an open source microservice management platform), automatic governance components, indicator collection and storage modules, data preprocessing modules, model inference modules, model training modules, and model deployment modules.

[0032] The indicator collection and storage module completes its work through open source Prometheus. By configuring Prometheus's prometheus.yml configuration file, it defines the target service exposed indicator endpoints that conform to the Prometheus format. In addition, the global configuration of the crawling interval can be set to 15s / time to obtain the latest performance data; the collected indicator data is efficiently stored in its built-in time series database. The collected data will pass through the data preprocessing module to perform operations such as cleaning and conversion on the raw data to obtain clean, consistent and easy-to-analyze data; then, based on the airline's business, the data features are selected to explore the trend, periodicity and seasonality in the data that may have a relevant impact on service traffic. Feature engineering is performed through the Pearson product-moment correlation coefficient. After data preprocessing, time series data is formed.

[0033] Exposing a monitoring endpoint is done by configuring an HTTP (HyperText Transfer Protocol) endpoint in the service. This endpoint returns the current service status and performance indicators, also known as access characteristics. Performance indicators are selected as request rate, response time, error rate (also known as access failure rate) and service uptime; for example, Figure 3 is a schematic diagram of the data collection and storage process provided in accordance with an embodiment of the present application, such as Figure 3 As shown in the following example, set the service port: determine a port number to expose the monitoring endpoint. This endpoint can be set to a non-service running port. Configure the Prometheus file: edit the prometheus.yml configuration file to define the targets and their endpoints. After saving the configuration file, start the Prometheus server. Prometheus periodically scrapes data from the service endpoint according to the configuration file settings and stores the collected data in Prometheus.

[0034] Step S102: input the access features into the target model to obtain the predicted access features of the target service processing access requests within the target time period adjacent to the preset period, wherein the target model is trained by multiple groups of training samples, each group of training samples includes the access features within the historical period and the access features within the next time period adjacent to the historical period, and the predicted access features include at least one of the following: request rate, request response time and request failure rate.

[0035] In step S102, the target model can be an LSTM (Long Short-Term Memory) model. When the collected historical access features reach a certain number after processing and meet the training cycle of the model, the data set is split in chronological order according to the ratio of 8:1:1, and constructed into a training set, a validation set, and a test set respectively. The training set is used for model training so that the model can learn the mapping relationship between historical access features and predicted access features, and continuously adjust the parameters of the model through the back propagation algorithm to minimize the loss function; the validation set is used to evaluate whether the setting of the model hyperparameters is reasonable, and the early stopping method is used to improve the generalization ability of the model; the test set is used to evaluate the performance of the model in real scenarios. After the evaluation is qualified, the model is deployed so that it can be used in a real production environment for real-time inference to obtain a forecast of service traffic for a period of time in the future. If the previous version of the model is already deployed in the current production environment, the model version is updated at 0 o'clock. After deployment, the predicted access features are obtained in real time through the target model.

[0036] Step S103 : collecting actual access features of access requests processed by the target service within the target time period, calculating the relative difference between the predicted access features and the actual access features, and obtaining a first difference value.

[0037] In step S103, the relative difference is calculated as follows: (|actual value-predicted value| / predicted value)*100%. The first difference value helps to evaluate the accuracy and stability of the model prediction. A lower first difference value means that the model's prediction results are closer to the actual service performance, thereby playing a more effective guiding role in the service automatic governance component. For example, if the first difference value between the predicted request rate and the actual request rate is low, then the circuit breaker degradation policy adjusted based on the prediction result can more accurately reflect the actual load of the service, thereby reducing unnecessary service interruptions or excessive traffic restrictions, improving user experience and reducing operation and maintenance costs. Calculations are performed regularly or after each prediction cycle by setting the corresponding program logic. This difference value will be used to adjust model parameters or trigger specific access request control policies to ensure that the model's predictive ability performs well in the real environment. In addition, recording the changing trend of the first difference value also helps to continuously monitor and optimize model performance, ensuring that its prediction accuracy is maintained or improved over time.

[0038] Step S104: determining an access request control policy based on a comparison result between the first difference value and a preset difference threshold, and executing the access request control policy on the target service.

[0039] In step S104, if the collected data has not reached the training cycle or the updated version, the model starts the process of reasoning on the data to obtain the predicted indicator results. The automatic governance component will extract the file from the storage point. When the automatic governance component identifies that the predicted access features and the actual access features match the access request control policy, the instruction is triggered and pushed to the configuration center. The business service uses the configuration center to maintain the dynamic current limiting configuration, and the current limiting instructions are dynamically loaded into the Sentinel (monitoring node) object to adjust the current limiting action. Dynamically set thresholds based on the prediction results, design and start the access request control policy of fuse or degradation, and ensure that measures can be taken in advance to prevent impending performance degradation or failure.

[0040] An access request control method provided in an embodiment of the present application collects access features of access requests processed by a target service within a preset period, wherein the preset period includes multiple time periods; inputs the access features into a target model to obtain predicted access features of access requests processed by the target service within a target time period adjacent to the preset period, wherein the target model is trained by multiple groups of training samples, each group of training samples includes access features within a historical period and access features within a next time period adjacent to the historical period, and the predicted access features include at least one of the following: request rate, request response time, and request failure rate; collects actual access features of access requests processed by the target service within the target time period, calculates the relative difference between the predicted access features and the actual access features, and obtains a first difference value; determines an access request control policy based on a comparison result between the first difference value and a preset difference threshold, and executes the access request control policy on the target service, thereby solving the problem of low response efficiency of the flow limiting policy of the service in the related art. By deploying the target model and predicting access characteristics in real time, setting thresholds based on the difference between the predicted results and the actual access characteristics, and designing and initiating access request control policies, we ensure that proactive measures can be taken to prevent impending performance degradation or failures. This reduces the need for manual intervention and significantly enhances the microservice architecture's ability to handle emergencies, enabling applications to operate continuously and stably in complex and changing environments. This, in turn, improves the responsiveness of the service's rate-limiting policies.

[0041] Determining a corresponding access request control policy based on a comparison result between the first difference value and different thresholds. Optionally, in the access request control method provided in an embodiment of the present application, determining the access request control policy based on a comparison result between the first difference value and a preset difference threshold includes: for each predicted access feature, when the first difference value is greater than or equal to the first difference threshold, determining closing the access link as the access request control policy, wherein the access link is a link for the user to access the target service; when the first difference value is less than the first difference threshold and greater than or equal to the second difference threshold, determining controlling the access frequency to be less than the first frequency threshold as the access request control policy, wherein the first difference threshold is greater than the second difference threshold, and the access frequency is the number of access requests to access the target service within the target duration; when the first difference value is less than the second difference threshold and greater than or equal to the third difference threshold, determining controlling the access frequency to be less than the second frequency threshold as the access request control policy, wherein the second frequency threshold is greater than the first frequency threshold, and the second difference threshold is greater than the third difference threshold; when the first difference value is less than the third difference value, determining not controlling the access frequency as the access request control policy.

[0042] In some embodiments, actual access characteristics reflect the true state of the current system or business and serve as the primary basis for decision-making. If actual access characteristics do not trigger fault tolerance conditions, indicating that the system is currently operating normally, predicted access characteristics can be used to assist in determining future trends in these characteristics. Experts pre-set difference thresholds based on their domain knowledge. Table 1 shows the structure of these difference thresholds.

[0043] Table 1

[0044] Light throttling threshold Medium throttling threshold Close new connections threshold Request rate 10% (third difference threshold) 20% (second difference threshold) 50% (first difference threshold) Response time 15% (third difference threshold) 30% (second difference threshold) 50% (first difference threshold) Error rate 5% (third difference threshold) 10% (second difference threshold) 20% (first difference threshold)

[0045] Logical judgment is performed based on the comparison of corresponding thresholds. If any indicator meets the conditions, the corresponding logical judgment branch is entered. If the threshold for closing new connections is exceeded, the new connection is closed, an emergency notification is issued to the operation and maintenance personnel, and the time, reason, and corresponding measures are recorded. If the moderate throttling conditions are met, moderate throttling measures are implemented, that is, controlling the access frequency to less than the first frequency threshold, and the time, reason, and corresponding measures are recorded. If the mild throttling conditions are met, mild throttling measures are implemented, that is, controlling the access frequency to less than the second frequency threshold, and the time, reason, and corresponding measures are recorded. If neither the actual value nor the predicted value meets the fault tolerance conditions, the next cycle is executed.

[0046] This embodiment uses adaptive thresholds to flexibly respond to changes and optimize resource allocation to avoid waste. This ensures stable service during high loads, minimizes service interruptions, improves user satisfaction, reduces manual intervention, and automates decision-making. It also enables continuous improvement and optimizes resource allocation. By setting thresholds based on scientific evidence, each decision is well-documented and easy to audit and manage.

[0047] This access request control strategy based on variance thresholds enables refined management of service traffic. In a microservices architecture, it helps quickly identify and respond to abnormal fluctuations in service load, proactively preventing overloads while minimizing the impact on normal services and maintaining overall system stability and robustness. Furthermore, by setting variance thresholds and control strategies at different levels, the system maximizes resource utilization and improves efficiency while maintaining service quality and performance.

[0048] If the access request control policy is executed, a prompt message needs to be issued in a timely manner. Optionally, in the access request control method provided in the embodiment of the present application, after executing the access request control policy on the target service, the method further includes: determining the execution time of the access request control policy and the target access feature that causes the execution of the access request control policy; obtaining the repair strategy associated with the target access feature, and issuing a prompt message, wherein the prompt message includes at least the execution time, the target access feature and the repair strategy, and the repair strategy is used to repair the target service.

[0049] In some embodiments, the system continuously monitors the access characteristics of the target service, including but not limited to request rate, response time, error rate, etc. Once it is found that the difference between these access characteristics and the predicted value exceeds a preset threshold (such as the first, second, and third difference thresholds), it means that the actual performance of the service deviates from expectations, and an access request control strategy may need to be adopted. When an abnormal target access characteristic is detected, the system needs to immediately calculate the optimal execution time of the access request control strategy. This is usually based on real-time evaluation and prediction of the service status (such as predicting future load changes) to ensure that the control strategy is executed at the most appropriate time to minimize the impact on normal services while effectively solving the problem. When triggering the control strategy, the system needs to identify the specific access characteristics that cause the anomaly. For example, if the request rate far exceeds the predicted value, then it may become a target feature that needs to be controlled.

[0050] Based on the target access characteristics, the system automatically or semi-automatically matches the corresponding remediation strategy. These strategies may include increasing server resources, optimizing service code, adjusting service architecture, etc. The specific strategy depends on the service type and the abnormal access characteristics. The execution time, the target access characteristics that caused the abnormality, and the matching remediation strategy are integrated into the prompt information. This ensures comprehensive information and clear operations. The prompt information is sent to the operation and maintenance personnel or relevant responsible persons, who will take action based on the information and implement the remediation strategy. The information can be sent through various channels, such as email, SMS, or directly displayed in the system interface, ensuring that the information reaches and is understood immediately.

[0051] The embodiment quickly responds to service exceptions through prompt information and provides specific operation guidance, helping the operation and maintenance team to timely and accurately solve problems, reducing service interruption time, and improving user experience. The implementation of the repair strategy is based on data-driven and intelligent decision-making, reducing the dependence on manual intervention and improving the efficiency and accuracy of service management.

[0052] After the access request control strategy is executed, if the monitored access feature meets the condition, the strategy can be recovered. Optionally, in the access request control method provided in the embodiment, after the access request control strategy is executed on the target service, the method further includes: monitoring the access feature of the target service within a preset time length to obtain a monitored access feature; calculating the relative difference between the monitored access feature and the predicted access feature to obtain a second difference value; determining whether the second difference value is less than or equal to a target difference threshold; in the case where the second difference value is greater than the target difference threshold, continuing to execute the step of executing the access request control strategy on the target service; and in the case where the second difference value is less than or equal to the target difference threshold, stopping executing the access request control strategy.

[0053] In some embodiments, Table 2 is a condition table for stopping executing the access request control strategy.

[0054] Table 2

[0055] Recycled access request control policy conditions Request rate When actual request rate falls below 90% of predicted value for more than 5 minutes Response time When average response time falls within 110% of predicted value for more than 5 minutes Error rate When error rate falls within 105% of predicted value for more than 5 minutes

[0056] If the recovery condition is not met, no processing is performed; if the recovery condition is met, the fault-tolerant configuration is deleted, and the next cycle is executed.

[0057] The embodiment dynamically adjusts the access request control strategy for the target service, ensuring that service management neither over-intervenes to cause resource waste or user experience degradation, nor under-intervenes to cause service instability or overload. The combination of the intelligence of the prediction model and the feedback of real-time monitoring forms a closed-loop service management mechanism, improving the efficiency and accuracy of service management.

[0058] To obtain the predicted access feature, a target model needs to be trained. Optionally, in the access request control method provided in the embodiment, the target model is obtained by the following method: obtaining historical access records of the target service, extracting access features in a plurality of historical periods from the historical access records, and extracting access features in a next historical period of each historical period from the historical access records; determining the access features in each historical period and the access features in the next historical period of the historical period as a set of training samples to obtain a plurality of sets of training samples; and training a neural network model based on the plurality of sets of training samples to obtain the target model.

[0059] In some embodiments, the data set is split in a time sequence according to a ratio of 8:1:1 to construct a training set, a validation set and a test set respectively. The training set is used for model training, for the model to learn the mapping relationship between features and labels, to continuously adjust the parameters of the model through the back propagation algorithm, so as to minimize the loss function; the validation set is used to evaluate whether the setting of the model hyperparameters is reasonable, and the early stopping method is used to improve the generalization ability of the model; the test set is used to evaluate the performance of the model in the real scene, and after the evaluation is qualified, the model is deployed to make real-time inference in the real production environment to obtain the predicted service traffic in the future period. If the current production environment already has a deployed previous version of the model, then the model version is updated at 0 o'clock. When the processed data reaches a certain amount and meets the training period of the model, the model introduces new data for offline optimization training, and when the training result is excellent, the developer updates the weight file and deploys it in the development environment at 0 o'clock. If the collected data does not reach the training period, the model directly performs inference service.

[0060] For example, in order to predict the changes of three specific features (request rate, response time, error rate) in the future time period, it is necessary to extract the past one day (96 time steps, 15 minutes per data point) from one month of data as input features. Each time point contains five feature values (request rate, response time, error rate, timestamp, weekday); a training sample is constructed, where the input feature dimension is (sample number, 96, 5), and the target output dimension is (sample number, 1, 3), representing the predicted values of the three features in the future 1 time step. The key pseudo code for LSTM data preparation is as follows. Where n_in = 96 represents the data of the past one day, n_out = 1 represents the data of one time step, and target_cols = [0, 1, 2] represents the prediction of the first three feature values.

[0061] The target model of the present embodiment can learn historical service access patterns and predict future access features based on these patterns. This enables service governance to become more proactive and intelligent, enabling it to anticipate possible access peaks or abnormal situations, thereby improving the stability and user experience of the service, while reducing the operation and maintenance cost.

[0062] The target model needs to be updated regularly. Optionally, in the access request control method provided in the present application, the method further comprises: reacquiring a plurality of updated training samples every target period, training the target model based on the plurality of updated training samples, and determining whether the updated target model is trained; in the case that the updated target model is trained, inputting the access feature into the updated target model to obtain the updated predicted access feature; in the case that the updated target model is not trained, performing the step of inputting the access feature into the target model to obtain the predicted access feature of the target service processing access request in the target period.

[0063] In some embodiments, a target period is defined, for example, daily, weekly or monthly, for re-evaluating and updating the prediction model. At the beginning or end of each target period, the historical access records of the target service are collected again, and the access features within multiple new historical periods and the access features within the next historical period of each historical period are extracted therefrom. The newly extracted access features are used to construct training samples, each sample containing the access features within a historical period as input, and the access features of the next period as output or label. Based on these new training samples, the neural network model is retrained. Adjusting the model parameters can also be fine-tuning the existing model to adapt to the new data and access patterns. Determine whether the updated target model has been trained. For example, the predetermined number of iterations is reached, the model performance indicators (such as accuracy or loss function value) meet the preset threshold, or the performance of the model no longer improves significantly (early stopping strategy).

[0064] If the updated target model is trained, the new model will be used as the updated target model. When predicting service access characteristics for the next target period, the access characteristics will be input into the updated model to obtain more accurate predicted access characteristics, thereby better guiding service governance and traffic control policies. If the updated target model is not trained, the system will continue to use the current target model to process prediction tasks until the new model is trained and verified.

[0065] This embodiment regularly checks and updates the target model to ensure that its predictive capabilities always keep pace with changes in service access patterns. This improves the system's adaptability and responsiveness, enabling it to more accurately predict service demand in the face of rapidly changing workload patterns or emergencies, and to promptly implement appropriate governance measures. This reduces downtime and recovery costs, optimizes resource allocation, improves user experience, reduces operation and maintenance costs, and achieves refined and intelligent service governance. Furthermore, through continuous model optimization, the system can continuously improve prediction accuracy and enhance overall business efficiency.

[0066] The access feature is screened out from a feature set. Optionally, in the access request control method provided in an embodiment of the present application, collecting the access features of the target service processing the access request within a preset period includes: obtaining the feature set of the target service processing the access request within the preset period, preprocessing the feature set to obtain multiple target features, wherein the preprocessing includes at least one of the following: removing duplicate data, outlier detection, missing value processing, data normalization and feature engineering; screening the baseline feature associated with the predicted access feature from the multiple target features, and for each target feature, calculating the similarity between the target feature and the baseline feature; when the similarity is within the preset similarity range, determining the target feature as the access feature.

[0067] In some embodiments, the data preprocessing module consists of three parts: data cleaning, data processing, and feature engineering, and the selected technology is dynamically adjusted by the model training results. Figure 4 is a schematic diagram of a data preprocessing module provided according to an embodiment of the present application, such as Figure 4 As shown, through the data cleaning step, outliers, duplicates, and missing values ​​are processed to obtain complete data. The data processing part encodes non-numeric features and processes the data into the same format and type, while standardizing the values. In addition, operations such as parsing and creating windows are performed for corresponding dates. The feature engineering part introduces time features by extracting holidays, weekdays, and internal airline schedules, and performs feature screening and correlation analysis. The calculation process of the correlation coefficient r, or similarity, can be expressed as:

[0068]

[0069] Where Cov(X, Y) is the covariance of X and Y; Var[X] and Var[Y] represent the variances of X and Y, respectively. X is the target feature and Y is the reference feature. The appropriate access feature set is finally screened out by comparing the similarity with the preset similarity range, and the output result is a data format that can be used for model training.

[0070] This embodiment improves the quality of the data through a preprocessing process to ensure that the most relevant and informative features are used during model training.

[0071] According to another embodiment of the present application, an optional access request control method is also provided. Figure 5 is a schematic diagram of an optional access request control method provided in an embodiment of the present application, such as Figure 5 As shown, the actual value of the indicator is obtained to determine whether the actual value of the indicator triggers the fault tolerance threshold; if triggered, determine whether it meets the conditions for closing new connections, moderate current limiting or mild current limiting. If the actual value of the indicator does not trigger the fault tolerance threshold, obtain the predicted value of the indicator, calculate the difference between the predicted value of the indicator and the actual value of the indicator, set the difference threshold, and determine whether it meets the conditions for closing new connections, moderate current limiting or mild current limiting based on the difference value and the difference threshold. If any condition is met, execute the corresponding strategy, record the time, reason and corresponding measures, and issue an emergency notification. After the strategy is executed, determine whether the actual value of the indicator meets the strategy recovery rules. If not, do not process it. If it is met, delete the fault tolerance configuration and execute the next cycle.

[0072] The optional access request control method of this embodiment, through the collaborative work of the above modules, can automatically issue and recover governance instructions when problems arise with airline website services, thereby improving overall service reliability. The LSTM model, a deep learning framework, is introduced to predict traffic metrics, develop circuit breaking and degradation rules, and predict future traffic metrics, enabling proactive action to reduce downtime and recovery costs. Adaptive thresholds enable flexible response to changes and optimize resource allocation to avoid waste. This ensures stable service under high load, minimizes service interruptions, and improves user satisfaction. Manual intervention is reduced, decision-making is automated, and continuous improvement optimizes resource allocation. Thresholds are set based on scientific evidence, ensuring that every decision is documented and easily auditable and manageable.

[0073] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0074] The present application also provides an access request control device. It should be noted that the access request control device of the present application can be used to execute the access request control method provided in the present application. The access request control device provided in the present application is introduced below.

[0075] Figure 6 Schematic diagram of a control device for access request provided in accordance with an embodiment of the present application. Figure 6 As shown, the device includes:

[0076] The collection unit 601 is configured to collect access characteristics of a target service processing access request within a preset period, wherein the preset period includes multiple time periods;

[0077] An input unit 602 is configured to input access features into a target model to obtain predicted access features of access requests processed by a target service within a target period adjacent to a preset period, wherein the target model is trained using multiple sets of training samples, each set of training samples including access features within a historical period and access features within a next period adjacent to the historical period, wherein the predicted access features include at least one of the following: request rate, request response time, and request failure rate;

[0078] A calculation unit 603 is configured to collect actual access characteristics of access requests processed by the target service within a target period, calculate a relative difference between the predicted access characteristics and the actual access characteristics, and obtain a first difference value;

[0079] The first determining unit 604 is configured to determine an access request control policy based on a comparison result between the first difference value and a preset difference threshold, and execute the access request control policy on the target service.

[0080] The access request control device provided by the embodiment of the present application collects access features of the target service processing access requests within a preset period through a collection unit 601, wherein the preset period includes multiple time periods; the input unit 602 inputs the access features into the target model to obtain predicted access features of the target service processing access requests within a target time period adjacent to the preset period, wherein the target model is trained by multiple groups of training samples, each group of training samples includes access features within a historical period and access features within a next time period adjacent to the historical period, and the predicted access features include at least one of the following: request rate, request response time, and request failure rate; the calculation unit 603 collects actual access features of the target service processing access requests within the target time period, and calculates The first difference value is obtained by comparing the predicted access characteristics with the actual access characteristics. A first determination unit 604 determines an access request control policy based on the comparison between the first difference value and a preset difference threshold, and then executes the access request control policy on the target service. This solves the problem of low response efficiency of service throttling policies in related technologies. By deploying a target model and predicting access characteristics in real time, setting a threshold based on the difference between the predicted result and the actual access characteristics, and designing and initiating an access request control policy, this ensures that proactive measures can be taken to prevent impending performance degradation or failure, reducing the need for manual intervention and significantly enhancing the microservice architecture's ability to handle emergencies, enabling applications to operate continuously and stably in complex and changing environments. This improves the response efficiency of service throttling policies.

[0081] Optionally, in the access request control device provided in the embodiment of the present application, the first determination unit 604 includes: a first determination module, used to, for each predicted access feature, determine closing the access link as the access request control strategy when the first difference value is greater than or equal to the first difference threshold, wherein the access link is the link for the user to access the target service; a second determination module, used to, when the first difference value is less than the first difference threshold and greater than or equal to the second difference threshold, determine controlling the access frequency to be less than the first frequency threshold as the access request control strategy, wherein the first difference threshold is greater than the second difference threshold, and the access frequency is the number of access requests to access the target service within the target duration; a third determination module, used to, when the first difference value is less than the second difference threshold and greater than or equal to the third difference threshold, determine controlling the access frequency to be less than the second frequency threshold as the access request control strategy, wherein the second frequency threshold is greater than the first frequency threshold, and the second difference threshold is greater than the third difference threshold; a fourth determination module, used to, when the first difference value is less than the third difference value, determine not controlling the access frequency as the access request control strategy.

[0082] Optionally, in the access request control device provided in the embodiment of the present application, the device also includes: a second determination unit, used to determine the execution time of the access request control policy and the target access feature that causes the execution of the access request control policy; a first acquisition unit, used to obtain the repair strategy associated with the target access feature, and issue a prompt message, wherein the prompt information includes at least the execution time, the target access feature and the repair strategy, and the repair strategy is used to repair the target service.

[0083] Optionally, in the access request control device provided in the embodiment of the present application, the device also includes: a monitoring unit, used to monitor the access characteristics of the target service within a preset time period to obtain the monitored access characteristics; a second calculation unit, used to calculate the relative difference between the monitored access characteristics and the predicted access characteristics to obtain a second difference value; a judgment unit, used to judge whether the second difference value is less than or equal to the target difference threshold; a first execution unit, used to continue to execute the step of executing the access request control policy on the target service when the second difference value is greater than the target difference threshold; and a stopping unit, used to stop executing the access request control policy when the second difference value is less than or equal to the target difference threshold.

[0084] Optionally, in the access request control device provided in the embodiment of the present application, the target model is trained in the following manner: obtaining historical access records of the target service, extracting access features within multiple historical periods from the historical access records, and extracting access features within the next historical period of each historical period from the historical access records; determining the access features within each historical period and the access features within the next historical period of the historical period as a group of training samples to obtain multiple groups of training samples; training a neural network model based on the multiple groups of training samples to obtain a target model.

[0085] Optionally, in the access request control device provided in the embodiment of the present application, the device also includes: a second acquisition unit, used to re-acquire the updated multiple sets of training samples every target period, train the target model based on the updated multiple sets of training samples, and determine whether the updated target model has been trained; a feature input unit, used to input the access features into the updated target model when the training of the updated target model is completed, to obtain updated predicted access features; a second execution unit, used to execute the step of inputting the access features into the target model when the training of the updated target model is not completed, to obtain predicted access features for access requests processed by the target service within the target time period.

[0086] Optionally, in the control device for access requests provided in an embodiment of the present application, the collection unit 601 includes: an acquisition module, used to obtain a feature set of the target service processing access requests within a preset period, and preprocess the feature set to obtain multiple target features, wherein the preprocessing includes at least one of the following: removal of duplicate data, outlier detection, missing value processing, data normalization and feature engineering; a screening module, used to screen baseline features associated with the predicted access features from multiple target features, and for each target feature, calculate the similarity between the target feature and the baseline feature; a fifth determination module, used to determine the target feature as an access feature when the similarity is within a preset similarity range.

[0087] The control device for access requests includes a processor and a memory. The above-mentioned acquisition unit 601, input unit 602, calculation unit 603 and first determination unit 604 are all stored in the memory as program units, and the processor executes the above-mentioned program units stored in the memory to realize corresponding functions.

[0088] The processor contains a kernel, which retrieves the corresponding program unit from the memory. One or more kernels can be set, and the response efficiency of the service's current limiting strategy can be improved by adjusting the kernel parameters.

[0089] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0090] An embodiment of the present invention provides a computer-readable storage medium storing a program, which implements a method for controlling an access request when the program is executed by a processor.

[0091] An embodiment of the present invention provides a processor, which is used to run a program, wherein a method for controlling an access request is executed when the program is running.

[0092] Figure 7 Schematic diagram of an electronic device according to an embodiment of the present application. Figure 7As shown, electronic device 701 includes a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, the following steps are performed: collecting access characteristics of access requests processed by a target service within a preset period, wherein the preset period includes multiple time periods; inputting the access characteristics into a target model to obtain predicted access characteristics of access requests processed by the target service within a target time period adjacent to the preset period, wherein the target model is trained using multiple sets of training samples, each set of training samples including access characteristics within a historical period and access characteristics within a next time period adjacent to the historical period, wherein the predicted access characteristics include at least one of the following: request rate, request response time, and request failure rate; collecting actual access characteristics of access requests processed by the target service within the target time period, calculating the relative difference between the predicted access characteristics and the actual access characteristics to obtain a first difference value; determining an access request control policy based on a comparison between the first difference value and a preset difference threshold, and executing the access request control policy on the target service. The device herein may be a server, a PC, a PAD, a mobile phone, etc.

[0093] The present application also provides a computer program product, which, when executed on a data processing device, is suitable for executing a program initialized with the following method steps: collecting access features of a target service processing access requests within a preset period, wherein the preset period includes multiple time periods; inputting the access features into a target model to obtain predicted access features of the target service processing access requests within a target time period adjacent to the preset period, wherein the target model is trained by multiple groups of training samples, each group of training samples includes access features within a historical period and access features within a next time period adjacent to the historical period, and the predicted access features include at least one of the following: request rate, request response time, and request failure rate; collecting actual access features of the target service processing access requests within the target time period, calculating the relative difference between the predicted access features and the actual access features, and obtaining a first difference value; determining an access request control policy based on a comparison result between the first difference value and a preset difference threshold, and executing the access request control policy on the target service.

[0094] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0095] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0096] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0097] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0098] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0099] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0100] Computer-readable media includes permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media (transitory media), such as modulated data signals and carrier waves.

[0101] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0102] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0103] The above are merely embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

Claims

1. A method for controlling access requests, characterized in that: include: Collecting access characteristics of target service processing access requests within a preset period, wherein the preset period includes multiple time periods; Inputting the access features into a target model to obtain predicted access features of the target service processing the access request in a target time period adjacent to the preset period, wherein the target model is trained by multiple sets of training samples, each set of training samples including access features in a historical period and access features in a next time period adjacent to the historical period, and the predicted access features including at least one of the following: request rate, request response time, and request failure rate; Collecting actual access characteristics of access requests processed by the target service within the target time period, calculating a relative difference between the predicted access characteristics and the actual access characteristics, and obtaining a first difference value; An access request control policy is determined based on a comparison result between the first difference value and a preset difference threshold, and the access request control policy is executed on the target service.

2. The method according to claim 1, characterized in that Determining the access request control policy based on the comparison result between the first difference value and a preset difference threshold includes: For each predicted access feature, when the first difference value is greater than or equal to a first difference threshold, determining closing the access link as the access request control policy, wherein the access link is a link for the user to access the target service; When the first difference value is less than the first difference threshold and greater than or equal to a second difference threshold, determining as the access request control policy that the access frequency is less than the first frequency threshold, wherein the first difference threshold is greater than the second difference threshold, and the access frequency is the number of access requests to the target service within a target duration; When the first difference value is less than the second difference threshold and greater than or equal to a third difference threshold, determining as the access request control policy that the access frequency is less than a second frequency threshold, wherein the second frequency threshold is greater than the first frequency threshold, and the second difference threshold is greater than the third difference threshold; In a case where the first difference value is smaller than the third difference value, not controlling the access frequency is determined as the access request control policy.

3. The method according to claim 2, characterized in that After executing the access request control policy on the target service, the method further includes: determining an execution time of the access request control policy and target access characteristics that require execution of the access request control policy; A repair strategy associated with the target access feature is obtained, and a prompt message is issued, wherein the prompt message at least includes the execution time, the target access feature, and the repair strategy, and the repair strategy is used to repair the target service.

4. The method according to claim 1, wherein After executing the access request control policy on the target service, the method further includes: Monitoring access characteristics of the target service within a preset time period to obtain monitored access characteristics; Calculating a relative difference between the monitored access feature and the predicted access feature to obtain a second difference value; Determining whether the second difference value is less than or equal to a target difference threshold; If the second difference value is greater than the target difference threshold, continue to execute the step of executing the access request control policy on the target service; When the second difference value is less than or equal to the target difference threshold, execution of the access request control policy is stopped.

5. The method according to claim 1, wherein The target model is trained in the following way: Obtaining historical access records of the target service, extracting access features in multiple historical periods from the historical access records, and extracting access features in the next historical period of each historical period from the historical access records; Determine the access features in each historical period and the access features in the next historical period of the historical period as a group of training samples, thereby obtaining multiple groups of training samples; A neural network model is trained based on the multiple groups of training samples to obtain the target model.

6. The method according to claim 5, characterized in that The method further comprises: Reacquiring multiple sets of updated training samples every target period, training the target model based on the multiple sets of updated training samples, and determining whether the training of the updated target model is complete; When the training of the updated target model is completed, the access features are input into the updated target model to obtain updated predicted access features; In the case that the updated target model has not been trained, the step of inputting the access features into the target model to obtain the predicted access features of the target service processing the access request within the target time period is performed.

7. The method according to claim 1, characterized in that The access characteristics of target service processing access requests collected within a preset period include: Obtaining a feature set of the target service processing access requests within the preset period, and preprocessing the feature set to obtain a plurality of target features, wherein the preprocessing includes at least one of the following: removing duplicate data, outlier detection, missing value processing, data normalization, and feature engineering; Selecting a reference feature associated with the predicted access feature from the plurality of target features, and calculating, for each target feature, a similarity between the target feature and the reference feature; When the similarity is within a preset similarity range, the target feature is determined as the access feature.

8. A control device for access request, characterized in that: include: a collection unit, configured to collect access characteristics of a target service processing access request within a preset period, wherein the preset period includes a plurality of time periods; an input unit, configured to input the access feature into a target model to obtain a predicted access feature of the target service processing the access request within a target period adjacent to the preset period, wherein the target model is trained by multiple sets of training samples, each set of training samples including access features within a historical period and access features within a next period adjacent to the historical period, the predicted access feature including at least one of the following: request rate, request response time, and request failure rate; a calculation unit, configured to collect actual access characteristics of access requests processed by the target service within the target time period, calculate a relative difference between the predicted access characteristics and the actual access characteristics, and obtain a first difference value; The first determining unit is configured to determine an access request control policy based on a comparison result between the first difference value and a preset difference threshold, and execute the access request control policy on the target service.

9. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the access request control method according to any one of claims 1 to 7 is implemented.

10. An electronic device, characterized in that: It includes one or more processors and a memory, the memory is used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the access request control method described in any one of claims 1 to 7.

Images