Interface traffic prediction method and device, electronic equipment and storage medium

Abstract

The invention discloses an interface traffic prediction method and device, electronic equipment and a storage medium, and relates to the technical field of traffic prediction, and the method comprises the steps: obtaining historical traffic data of an interface, and extracting historical multi-modal parameters in the historical traffic data, the historical multi-modal parameters at least comprise a historical time sequence, a historical request quantity, a request path, a request parameter, a response parameter and response time; and inputting the historical multi-modal parameters into a preset recurrent neural network to obtain a predicted request quantity of the interface at a preset moment. According to the method and the device, various factors, such as a historical time sequence, a historical request quantity, a request path, a request parameter, a response parameter and response time, which influence future traffic of an interface are comprehensively considered by adopting modes of historical traffic data acquisition, multi-modal parameter extraction and RNN processing of multi-modal parameters; therefore, the effect of accurately predicting the future flow of the interface is achieved.

Description

Technical Field

[0001] The present application relates to the technical field of traffic prediction, and in particular to an interface traffic prediction method, device, electronic device and storage medium. Background Art

[0002] With the rapid development of Internet technology, the demand for network services continues to grow. Accurately predicting the traffic demand of API (Application Programming Interface) in order to reasonably allocate resources and improve service quality and user experience has become an important topic of concern in the industry. By analyzing historical traffic data and using machine learning, deep learning and other algorithms to establish a traffic prediction model, it is possible to effectively predict the traffic demand of the interface for a period of time in the future, thereby realizing dynamic scheduling and optimal configuration of resources, and improving network service performance and resource utilization. However, in the scheme adopted by the existing traffic prediction technology, the collected training data is the historical time series and the request volume, that is, the training data of the scheme is single-mode data, in which only the interface corresponds to the request volume at a certain point in time, and the future traffic prediction of the interface based on the single-mode data is often less accurate. Therefore, the existing traffic prediction technology does not fully consider the various factors that affect the future traffic of the interface, and it is difficult to accurately predict the interface traffic.

[0003] Therefore, how to accurately predict the future traffic of an interface is a problem that needs to be solved urgently. Summary of the invention

[0004] The main purpose of this application is to provide an interface traffic prediction method, device, electronic device and storage medium, aiming to solve the technical problem of how to accurately predict the future traffic of an interface.

[0005] To achieve the above object, the present application proposes an interface traffic prediction method, the method comprising: Obtaining historical traffic data of the interface, and extracting historical multimodal parameters in the historical traffic data, wherein the historical multimodal parameters at least include historical time series, historical request volume, request path, request parameters, response parameters, and response time; The historical multimodal parameters are input into a preset recursive neural network to obtain the predicted request volume of the interface at a preset time.

[0006] In one embodiment, after the step of inputting the historical multimodal parameters into a preset recursive neural network to obtain the predicted request volume of the interface at a preset time, the step further includes: Monitoring the actual request volume of the interface at a preset time, and comparing the actual request volume with the predicted request volume to obtain a comparison result; The model parameters of the preset recursive neural network are adjusted based on the comparison result.

[0007] In one embodiment, the step of inputting the historical multimodal parameters into a preset recursive neural network to obtain the predicted request volume of the interface at a preset time includes: Using the last predicted request amount output by the preset recursive neural network as the memory request amount; The historical multimodal parameters and the memory request amount are input into the preset recursive neural network, and the predicted request amount is calculated in the preset recursive neural network according to the memory request amount and the historical multimodal parameters.

[0008] In one embodiment, the step of calculating the predicted request amount in the preset recursive neural network according to the remembered request amount and the historical multimodal parameter comprises: Vectorizing the historical multimodal parameters to obtain a historical multimodal vector; Multiplying a first model parameter of the preset recursive neural network by the historical multimodal vector through the preset recursive neural network to obtain a first multiplication result; multiplying the second model parameter of the preset recursive neural network by the memory request amount through the preset recursive neural network to obtain a second multiplication result; The predicted request amount is obtained by taking the hyperbolic tangent value of the sum of the first multiplication result and the second multiplication result through the preset recursive neural network, and adding the hyperbolic tangent value to the third model parameter of the preset recursive neural network.

[0009] In one embodiment, the model parameters include a first model parameter, a second model parameter, and a third model parameter, and the step of comparing the actual request amount with the predicted request amount to obtain a comparison result includes: Comparing the actual request amount with the predicted request amount to obtain a difference value between the actual request amount and the predicted request amount; Calculating a mean square error of the preset recursive neural network based on the difference value; A first gradient of the first model parameter, a second gradient of the second model parameter, and a third gradient of the third model parameter are calculated according to the mean square error to obtain the comparison result.

[0010] In one embodiment, the interface traffic prediction method further includes: Determine a target request quantity threshold value among a plurality of preset request quantity threshold values ​​based on the predicted request quantity, and obtain a scheduling instruction corresponding to the target request quantity threshold value, wherein, when the predicted request quantity is greater than the preset expansion threshold value, the target request quantity threshold value is a preset request quantity threshold value that is adjacent to the predicted request quantity and smaller than the predicted request quantity, and when the predicted request quantity is less than the preset reduction threshold value, the target request quantity threshold value is a preset request quantity threshold value that is adjacent to the predicted request quantity and larger than the predicted request quantity; The scheduling instruction is sent to the cluster to be scheduled, so that the cluster to be scheduled adjusts the capacity of the service instances in the cluster to be scheduled based on the scheduling instruction.

[0011] In one embodiment, the cluster to be scheduled includes a K8s cluster, and the step of sending the scheduling instruction to the cluster to be scheduled so that the cluster to be scheduled adjusts the capacity of the service instances in the cluster to be scheduled based on the scheduling instruction includes: When the predicted request amount is greater than a preset expansion threshold, the scheduling instruction is sent to the K8s cluster, so that the K8s cluster increases the number of service instances in the K8s cluster and / or increases the resources allocated to the service instances based on the scheduling instruction; When the predicted request amount is less than the preset shrinking threshold, the scheduling instruction is sent to the K8s cluster so that the K8s cluster reduces the number of service instances in the K8s cluster and / or reduces the resources allocated to the service instances based on the scheduling instruction.

[0012] In addition, to achieve the above purpose, the present application also proposes an interface traffic prediction device, the interface traffic prediction device comprising: An acquisition module, used to acquire historical traffic data of an interface and extract historical multimodal parameters from the historical traffic data, wherein the historical multimodal parameters at least include historical time series, historical request volume, request path, request parameters, response parameters and response time; The output module is used to input the historical multimodal parameters into a preset recursive neural network and output the predicted request volume of the interface at a preset time.

[0013] In addition, to achieve the above-mentioned purpose, the present application also proposes an electronic device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the interface traffic prediction method as described above.

[0014] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by the processor, the steps of the interface traffic prediction method described above are implemented.

[0015] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, the steps of the interface traffic prediction method described above are implemented.

[0016] One or more technical solutions proposed in this application have at least the following technical effects: The present application first obtains historical traffic data of the interface and extracts historical multimodal parameters from the historical traffic data, so as to extract key features of traffic changes, namely historical multimodal parameters, by collecting and analyzing the historical traffic data of the interface, thereby providing a reliable data basis and key influencing factors for subsequent traffic prediction; the historical multimodal parameters are input into a preset recursive neural network to obtain the predicted request volume of the interface at a preset time, and the learning ability of the recursive neural network is combined with the historical multimodal parameters to predict the future request volume of each interface, thereby achieving accurate prediction of the traffic of each interface in advance.

[0017] In summary, this application comprehensively considers various factors affecting the future traffic of the interface, such as historical time series, historical request volume, request path, request parameters, response parameters and response time, by adopting the method of historical traffic data collection, multimodal parameter extraction and RNN (Recursive Neural Network) processing multimodal parameters, thereby achieving the effect of accurately predicting the future traffic of the interface. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0019] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0020] Figure 1 A flow chart of the first embodiment of the method for predicting interface traffic in this application; Figure 2 A flow chart of the second embodiment of the method for predicting interface traffic in this application; Figure 3 A schematic diagram of the scheduling process of the interface traffic prediction method provided in Example 2 of the present application; Figure 4 A brief flowchart of the interface traffic prediction method provided in Example 2 of the present application; Figure 5 A schematic diagram of an RNN model of an interface traffic prediction method provided in Example 2 of the present application; Figure 6 This is a schematic diagram of the module structure of the interface traffic prediction device according to an embodiment of the present application; Figure 7 This is a schematic diagram of the device structure of the hardware operating environment involved in the interface traffic prediction method in the embodiment of the present application.

[0021] The purpose, features and advantages of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0022] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0023] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0024] The main solution of the embodiment of the present application is: obtaining historical traffic data of the interface, and extracting historical multimodal parameters in the historical traffic data, wherein the historical multimodal parameters at least include historical time series, historical request volume, request path, request parameters, response parameters and response time; inputting the historical multimodal parameters into a preset recursive neural network, and outputting the predicted request volume of the interface at a preset time.

[0025] In the scheme adopted by the existing traffic prediction technology, the collected training data are historical time series and request volume, that is, the training data of the scheme is single-mode data, in which only the request volume corresponding to the interface at a certain point in time is included, and the prediction of the future traffic of the interface based on the single-mode data is often less accurate. Therefore, the existing traffic prediction technology does not fully consider the various factors that affect the future traffic of the interface, and thus it is difficult to accurately predict the traffic of the interface. Therefore, how to accurately predict the future traffic of the interface is a problem that needs to be solved urgently.

[0026] The present application provides a solution, which comprehensively considers various factors affecting the future traffic of the interface, such as historical time series, historical request volume, request path, request parameters, response parameters and response time, by adopting the methods of historical traffic data collection, multimodal parameter extraction and RNN processing of multimodal parameters, so as to achieve the effect of accurately predicting the future traffic of the interface.

[0027] It should be noted that the execution subject of this embodiment may be a computing service device with data processing, network communication and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or an electronic device capable of realizing the above functions, etc. The following takes an electronic device as an example to illustrate this embodiment and the following embodiments.

[0028] Based on this, the present application embodiment provides an interface traffic prediction method, referring to Figure 1 , Figure 1 This is a flow chart of the first embodiment of the interface traffic prediction method of the present application.

[0029] In this embodiment, the interface traffic prediction method includes steps S10 to S20: Step S10, obtaining historical traffic data of the interface, and extracting historical multimodal parameters in the historical traffic data, wherein the historical multimodal parameters at least include historical time series, historical request volume, request path, request parameters, response parameters and response time; It should be noted that historical traffic data refers to the traffic record of the interface within a preset period of time in the past; historical multimodal parameters refer to the key features extracted from the historical traffic data that can reflect the interface traffic, which may include the historical request volume, request parameters, request path, response parameters, response time, etc. of the interface; historical time series refers to the time period for the interface to record traffic transmission; historical request volume refers to the number of requests received by the interface within a specific time period in the past; request path refers to the type identifier of the interface called when initiating a request; request parameters refer to the parameters attached when initiating a request, which are usually used for query conditions, data filtering, passing variables, etc.; response parameters refer to the results and related data returned in response to the request; response time refers to the time from initiating a request to returning a response.

[0030] It is understandable that, in the process of predicting the future traffic of the interface, the future request volume of the interface is not only related to the historical request volume, but also to parameters such as the parameters of the historical requests, response parameters and response time. Relying only on a single data will lead to prediction bias and lack of robustness. Therefore, performing step S10 can avoid the problems of excessive deviation and low robustness in predicting the future traffic of the interface, thereby ensuring that the RNN can capture the complex changes and long-term trends of the traffic, thereby improving the accuracy and stability of traffic prediction.

[0031] Exemplarily, historical traffic data of the API is obtained, and the request volume, request parameters, request path, response parameters, and response time corresponding to each historical time series in the historical traffic data are determined.

[0032] Step S20, inputting the historical multimodal parameters into a preset recursive neural network to obtain a predicted request volume of the interface at a preset time; It should be noted that the preset recursive neural network refers to a special series of neural networks, which is mainly used to process sequence data. The sequence data often has temporal correlation, that is, the output of the network at a certain moment is not only related to the input at the current moment, but also related to the output at a certain moment or several moments before; the predicted request volume refers to the number of requests that the interface may make in a certain time period in the future.

[0033] It is understandable that relying solely on traditional statistical methods or simple time series analysis of the interface's historical traffic data may lead to inaccurate predictions, especially when faced with nonlinear and non-stationary traffic data. Therefore, step S20 is performed. Through the sequence learning ability of the recursive neural network, the long-term dependencies and dynamic changes of the historical traffic data can be captured, thereby providing a more accurate and reliable interface traffic prediction.

[0034] In a feasible implementation, step S20 may include steps S21-S22: Step S21, using the last predicted request amount output by the preset recursive neural network as the memory request amount; It should be noted that the previous predicted request amount refers to the predicted request amount predicted by the RNN in the previous time step; the memorized request amount refers to using the previous predicted request amount as part of the RNN input of the current time step, which is used for the RNN to memorize the previous traffic status.

[0035] It is understandable that since existing traffic prediction technologies often ignore the impact of historical prediction results of neural networks on current predictions, step S21 is performed to avoid ignoring the impact of instance prediction results on current predictions, thereby maintaining the continuity of the prediction and the association with the context, making the prediction results more in line with the actual situation.

[0036] Step S22, inputting the historical multimodal parameters and the memorized request amount into the preset recursive neural network, and calculating the predicted request amount in the preset recursive neural network according to the memorized request amount and the historical multimodal parameters.

[0037] It is understandable that since only relying on the traffic data at the current moment may lead to short-sighted prediction problems, step S22 is performed. By adopting RNN to comprehensively consider the historical traffic trends and the most recent traffic status, the request volume of each interface can be more accurately predicted, thereby improving the accuracy of the traffic prediction method in predicting the interface request volume and the adaptability to traffic fluctuations.

[0038] In this embodiment, by utilizing the memory characteristics of RNN, the previous predicted request volume is maintained in the request volume prediction for the time series, and the current predicted request volume is calculated in combination with historical multimodal parameters, thereby improving the continuity and accuracy of the predicted traffic and making the traffic prediction more in line with the actual dynamic changes of the time series.

[0039] In a feasible implementation manner, the step of calculating the predicted request amount in the preset recursive neural network according to the remembered request amount and the historical multimodal parameter in step S22 may include steps S221 to S224: Step S221, vectorizing the historical multimodal parameters to obtain a historical multimodal vector; It should be noted that the historical multimodal vector refers to the numerical vector form in which the extracted historical multimodal parameters are converted, and this vector form is convenient for RNN processing.

[0040] It can be understood that since it is difficult for RNN to directly process unstructured data, step S221 is performed to improve the data processing efficiency of RNN by converting the parameters into a vector format acceptable to RNN.

[0041] Step S222, multiplying a first model parameter of the preset recursive neural network by the historical multimodal vector through the preset recursive neural network to obtain a first multiplication result; It is understandable that, since it is often difficult to effectively extract the characteristics of interface traffic when using neural networks for traffic prediction, step S222 is performed to preliminarily extract and weight historical multimodal features through linear transformation, providing an effective data basis for subsequent nonlinear processing.

[0042] Step S223, multiplying the second model parameter of the preset recursive neural network by the memory request amount through the preset recursive neural network to obtain a second multiplication result; It is understandable that, since the influence of the historical prediction results of the neural network on the current prediction is often ignored in the process of using the neural network for traffic prediction, step S223 is performed to combine the historical prediction information of the neural network so that the neural network can consider the continuity of the time series, thereby improving the accuracy of the traffic prediction method in predicting the interface request volume.

[0043] Step S224: Obtain the hyperbolic tangent value of the sum of the first multiplication result and the second multiplication result through the preset recursive neural network, and add the hyperbolic tangent value to the third model parameter of the preset recursive neural network to obtain the predicted request amount.

[0044] It is understandable that in the process of using neural networks for traffic prediction, the prediction results may be too linear and thus unable to capture the complex relationship of interface traffic, so step S224 is performed to introduce nonlinear factors through a nonlinear activation function (hyperbolic tangent function) and combine it with a bias term (the third model parameter) to improve the adaptability of the traffic prediction method to traffic fluctuations.

[0045] Exemplarily, the predicted request volume h is calculated by the following formula: t :

[0046] Among them, U is the first model parameter, x t is the historical multimodal vector, W is the second model parameter, h t-1 is the memory request amount, C is the third model parameter, and tanh is the hyperbolic tangent function.

[0047] In this implementation, the parameters input to the RNN are vectorized and combined with the internal model parameters of the RNN. The complex dynamic relationship of the traffic data is captured using linear transformation and nonlinear activation functions, and the predicted request volume is finally obtained. This improves the accuracy of predicting future traffic of the interface while taking into account time series dependencies.

[0048] This embodiment provides an interface traffic prediction method, which comprehensively considers various factors affecting the future traffic of the interface, such as historical time series, historical request volume, request path, request parameters, response parameters and response time, by adopting historical traffic data collection, multimodal parameter extraction and RNN processing of multimodal parameters, thereby achieving the effect of accurately predicting the future traffic of the interface.

[0049] In a feasible implementation manner, step S20 further includes steps S31-S32: Step S31, monitoring the actual request volume of the interface at a preset time, and comparing the actual request volume with the predicted request volume to obtain a comparison result; It should be noted that the actual request volume refers to the number of requests actually made by the interface within a specific time period.

[0050] It is understandable that since RNN may produce prediction deviations due to improper initialization parameters or changes in traffic data, resulting in inaccurate predicted request volumes, step S31 is performed to provide an effective adjustment basis for adjusting the RNN model parameters by comparing the actual request volumes with the predicted request volumes.

[0051] In a feasible implementation manner, the step of comparing the actual request amount with the predicted request amount to obtain a comparison result in step S31 may include steps S311 to S313: Step S311, comparing the actual request amount with the predicted request amount to obtain a difference value between the actual request amount and the predicted request amount; It should be noted that the difference value refers to the difference between the actual request amount and the predicted request amount, and the difference quantifies the accuracy of the prediction.

[0052] It can be understood that since the adjustment of the RNN model parameters needs to be determined based on the prediction error, step S311 is performed to avoid the problem of being unable to quantify the prediction error, provide a direct indicator for measuring the prediction accuracy of the RNN, and provide an effective data basis for adjusting the RNN model parameters.

[0053] Step S312, calculating the mean square error of the preset recursive neural network based on the difference value; It should be noted that the mean square error refers to the average value of the squared difference values, which measures the overall level of the difference between the predicted request volume and the actual request volume.

[0054] It is understandable that since it is impossible to comprehensively evaluate the RNN prediction performance based on only a single difference value, step S312 is performed to obtain a comprehensive error metric by calculating the mean square error, thereby achieving the effect of evaluating the overall prediction performance of the RNN and providing an effective data basis for adjusting the model parameters of the RNN.

[0055] Step S313: Calculate the first gradient of the first model parameter, the second gradient of the second model parameter, and the third gradient of the third model parameter according to the mean square error to obtain the comparison result.

[0056] It should be noted that the first gradient refers to the gradient of the mean square error corresponding to the first model parameter; the second gradient refers to the gradient of the mean square error corresponding to the second model parameter; the third gradient refers to the gradient of the mean square error corresponding to the third model parameter; the gradient refers to the partial derivative of the mean square error, which indicates the direction in which the mean square error rises fastest in the parameter space.

[0057] It is understandable that, since it is often unclear how to adjust the model parameters to reduce errors during the process of adjusting the model parameters, step S313 is performed to calculate the gradient of the model parameters, thereby providing an effective data basis for guiding the adjustment of the model parameters based on subsequent optimization algorithms such as the gradient descent algorithm to reduce the prediction error.

[0058] In this implementation, by accurately quantifying the prediction error, evaluating the RNN prediction performance by the whole population, and using gradient information to guide the adjustment of model parameters, the deviation of RNN prediction is reduced, and the prediction accuracy of RNN for interface traffic and the generalization ability of RNN are improved.

[0059] Step S32: adjusting the model parameters of the preset recursive neural network based on the comparison result.

[0060] It should be noted that model parameters refer to adjustable parameters used in RNN to calculate the predicted request amount, including weights (such as the first model parameter and the second model parameter) and biases (such as the third model parameter).

[0061] It is understandable that due to the problem of inaccurate predicted request volume during actual use of RNN, step S32 is performed to ensure the optimization of RNN model parameters by adjusting the RNN model parameters in real time, thereby improving the accuracy of the prediction and the adaptability of the model, thereby making the prediction results closer to the actual traffic situation.

[0062] In a feasible implementation manner, the model parameters include a first model parameter, a second model parameter and a third model parameter, and the step of adjusting the model parameters of the preset recursive neural network based on the comparison result in step S32 may include step S321: Step S321, when the comparison result is the first gradient of the first model parameter, the second gradient of the second model parameter and the third gradient of the third model parameter, by adopting a gradient descent algorithm in the preset recursive neural network, the first model parameter is adjusted according to the first gradient, the second model parameter is adjusted according to the second gradient, and the third model parameter is adjusted according to the third gradient.

[0063] It should be noted that the first model parameter refers to the weight matrix in the RNN, the second model parameter refers to another weight matrix in the RNN, and the third model parameter refers to the bias term in the RNN.

[0064] It is understandable that since there may be problems of blind adjustment or failure to improve the prediction accuracy after adjustment in the process of adjusting the RNN model parameters, step S321 is performed, and the gradient descent optimization algorithm is adopted by the preset optimizer to achieve the effect of accurately adjusting the model parameters to improve the model prediction accuracy.

[0065] For example, when the comparison result is U, i.e., the first model parameter ∇ U That is, the first gradient, W is the second model parameter ∇ W That is, the second gradient, C is the third model parameter ∇ C That is, in the case of the third gradient, based on the preset optimizer, the preset learning rate of RNN is subtracted from U and ∇ U The product of ∇ and ∇ is used to obtain the adjusted U, which is based on W minus the preset learning rate of RNN and ∇ W The product of ∇ is used to obtain the adjusted W, which is based on C minus the preset learning rate of RNN and ∇C The product of is the adjusted C.

[0066] In this implementation, by adjusting the first model parameter (weight matrix), the second model parameter (another weight matrix) and the third model parameter (bias term) of the RNN according to the gradient of the mean square error with respect to the model parameters, it is ensured that the model can effectively learn from the prediction deviation and continuously improve, thereby improving the accuracy of the predicted traffic and the performance of the model.

[0067] This embodiment provides an interface traffic prediction method, which realizes real-time monitoring and prediction of changes in interface traffic by adopting historical traffic data collection, multimodal parameter extraction, RNN training and dynamic adjustment of model parameters, ensuring that the RNN can continuously learn and adjust parameters according to actual traffic data, thereby achieving the effect of accurately predicting future interface traffic.

[0068] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those in the above-mentioned embodiment 1 can be referred to the above introduction, and will not be repeated in the following. Figure 2 , Figure 2 This is a flow chart of the second embodiment of the interface traffic prediction method of the present application.

[0069] The interface traffic prediction method further includes steps S01-S02: Step S01, determining a target request quantity threshold among a plurality of preset request quantity thresholds based on the predicted request quantity, and obtaining a scheduling instruction corresponding to the target request quantity threshold, wherein, when the predicted request quantity is greater than the preset expansion threshold, the target request quantity threshold is a preset request quantity threshold adjacent to the predicted request quantity and smaller than the predicted request quantity, and when the predicted request quantity is less than the preset reduction threshold, the target request quantity threshold is a preset request quantity threshold adjacent to the predicted request quantity and larger than the predicted request quantity; It should be noted that the preset request volume threshold refers to a series of pre-set request volume levels, which are used to determine the appropriate triggering resource scheduling action; the target request volume threshold refers to the threshold that is closest to the predicted request volume and is less than or greater than the predicted request volume among many preset request volume thresholds; the scheduling instruction refers to the instruction used to guide resource scheduling for resource allocation or adjustment; the preset expansion threshold refers to a fixed value pre-set to indicate that the predicted request volume is too large and the cluster service needs to be expanded; the preset reduction threshold refers to a fixed value pre-set to indicate that the predicted request volume is too small and the cluster service needs to be reduced.

[0070] In addition, it should be noted that, when the predicted request volume is less than or equal to the preset expansion threshold, and the predicted request volume is greater than or equal to the preset reduction threshold, the target request volume threshold does not exist, and there is no need to adjust the capacity of the service instances in the cluster to be scheduled, that is, the step of sending the scheduling instruction to the cluster to be scheduled so that the cluster to be scheduled adjusts the capacity of the service instances in the cluster to be scheduled based on the scheduling instruction is not executed.

[0071] It is understandable that since the existing technical solutions need to manually perform dynamic expansion or reduction of the service after predicting the future traffic trend of the interface, resulting in inefficient capacity adjustment and possible problems with inappropriate resource scheduling timing, step S01 is performed to automatically and accurately determine the timing of obtaining the scheduling instruction based on the predicted request volume, thereby preparing resources in advance according to the predicted traffic demand, ensuring that the service can smoothly and reasonably respond to the upcoming request volume.

[0072] Step S02: sending the scheduling instruction to the cluster to be scheduled, so that the cluster to be scheduled adjusts the capacity of the service instances in the cluster to be scheduled based on the scheduling instruction.

[0073] It should be noted that the cluster to be scheduled refers to a computing resource cluster that needs to be scheduled and managed; the service instance refers to a specific application running on the cluster, which is responsible for processing interface requests.

[0074] It is understandable that since the services in the cluster may be unable to handle peak requests due to insufficient resources, or may waste resources due to excessive resources, step S02 is performed to dynamically adjust the capacity of the service instances to ensure that the service instances can meet the predicted traffic demand while saving resources, thereby improving the availability and reliability of the service instances.

[0075] In a feasible implementation manner, the cluster to be scheduled includes a K8s cluster, and step S02 may include step A021: Step A201, when the predicted request amount is greater than the preset expansion threshold, the scheduling instruction is sent to the K8s cluster so that the K8s cluster increases the number of service instances in the K8s cluster and / or increases the resources allocated to the service instances based on the scheduling instruction.

[0076] It should be noted that the K8s cluster refers to a containerized application management platform built on Kubernetes (K8s) technology, which can automatically deploy, expand and manage containerized applications.

[0077] For example, please refer to Figure 3, the user pre-configures different scheduling strategies based on each preset request volume threshold, and each scheduling strategy corresponds to a specific scheduling instruction, such as: when the predicted request volume is greater than 1000 times / second, it corresponds to the K8s instruction, i.e., the scheduling instruction, to double the capacity of the service instance; when the predicted request volume is greater than 2000 times / second, it corresponds to the K8s instruction to double the capacity of the service instance. When the K8s instruction is determined based on the prediction result, i.e., the predicted request volume, the K8s instruction is sent to the K8s cluster so that the K8s cluster automatically schedules the service instance for expansion.

[0078] In this implementation, by automatically increasing the number of service instances and / or automatically increasing the resources that can be called by service instances based on scheduling instructions, the problems of inefficiency and response delay caused by manual adjustment of resources are avoided, and the K8s cluster is ensured to automatically increase service instances in advance according to predicted traffic demand, thereby responding to traffic peaks in a timely manner and maintaining the stability and availability of service instances.

[0079] In another feasible implementation, the cluster to be scheduled includes a K8s cluster, and step S02 may include step B021: Step B021, when the predicted request amount is less than the preset shrinking threshold, the scheduling instruction is sent to the K8s cluster so that the K8s cluster reduces the number of service instances in the K8s cluster and / or reduces the resources allocated to the service instances based on the scheduling instruction.

[0080] It should be noted that resources refer to the extent to which a service instance can call upon system resources (such as CPU, memory, etc.).

[0081] Exemplarily, the user pre-configures different scheduling strategies based on each preset request volume threshold, and each scheduling strategy corresponds to a specific scheduling instruction, such as: when the predicted request volume is less than 200 times / second, it corresponds to the K8s instruction, i.e., the scheduling instruction, to shrink the service instance by one time, and when the predicted request volume is less than 100 times / second, it corresponds to the K8s instruction to shrink the service instance by two times. When the K8s instruction is determined based on the prediction result, i.e., the predicted request volume, the K8s instruction is sent to the K8s cluster so that the K8s cluster automatically schedules the service instance to shrink.

[0082] In this implementation, by automatically reducing the amount of callable resources of the service instance based on scheduling instructions, the problem of resource waste caused by excessive resource occupation by the service instance is avoided, and the K8s cluster is ensured to automatically reduce the amount of callable resources in advance according to predicted traffic demand, thereby saving occupied resources in time and improving the performance and power consumption of the cluster where the service instance is located.

[0083] In this embodiment, by adopting an automatic capacity adjustment method based on the predicted request volume, the service instance is automatically expanded or reduced according to the predicted traffic demand, ensuring that the service instances in the K8s cluster can respond to traffic changes in a timely manner and maintain the stable operation and high availability of the service instances.

[0084] For example, to help understand the implementation process of the interface traffic prediction method obtained by combining this embodiment with the above-mentioned embodiment 1, please refer to Figure 4 , Figure 4 A brief flow chart of an interface traffic prediction method is provided, specifically: First, historical traffic data of multiple interfaces are obtained, and historical multimodal parameters are extracted from the historical traffic data as shown in Table 1:

[0085] Table 1 In Table 1, the historical multimodal parameters are historical time series, request volume (i.e., historical request volume), request path, request parameters, response parameters, and response time. These parameters are vectorized and input into the API traffic prediction model, i.e., the preset RNN, to obtain the predicted request volume of the interface, and the predicted request volume is input into the automatic scheduler, so that the automatic scheduler sends a scheduling instruction based on the predicted request volume, and sends the scheduling instruction to the K8s cluster to complete the capacity adjustment of the service instance in the K8s cluster.

[0086] For further information, please refer to Figure 5 , Figure 5 A schematic diagram of an RNN model for an interface traffic prediction method is provided. In the figure, the contents of the dashed boxes A on both sides are consistent with the contents of the middle dashed box. , , The RNN model input for the previous moment, current moment, and next moment, , , The RNN model output at the previous moment, the current moment, and the next moment, wherein, in the first embodiment and / or the second embodiment, is the historical multimodal parameter, is the predicted request volume of the interface. The RNN model proposed in this embodiment includes 24 cells, i.e., dotted boxes, corresponding to 24 hours in a day. Figure 5 The figure shows the connection of three cells. Through this cascade design, the current output Not only with the current input It is related to the output of the previous moment. It is related to and also affects the output at the next moment .

[0087] It should be noted that the above examples are only used to understand the present application and do not constitute a limitation on the interface traffic prediction method of the present application. More simple transformations based on this technical concept are all within the scope of protection of the present application.

[0088] This application also provides an interface traffic prediction device, please refer to Figure 6 , the interface traffic prediction device comprises: An acquisition module 10 is used to acquire historical traffic data of an interface and extract historical multimodal parameters from the historical traffic data, wherein the historical multimodal parameters at least include historical time series, historical request volume, request path, request parameters, response parameters and response time; The output module 20 is used to input the historical multimodal parameters into a preset recursive neural network to obtain the predicted request volume of the interface at a preset time.

[0089] Optionally, the adjustment module 30 in the interface traffic prediction device is used to: Using the last predicted request amount output by the preset recursive neural network as the memory request amount; The predicted request amount is calculated in the preset recursive neural network according to the remembered request amount and the historical multimodal parameters.

[0090] Optionally, the output module 20 is further used for: Using the last predicted request amount output by the preset recursive neural network as the memory request amount; The historical multimodal parameters and the memory request amount are input into the preset recursive neural network, and the predicted request amount is calculated in the preset recursive neural network according to the memory request amount and the historical multimodal parameters.

[0091] Optionally, the output module 20 is further used for: Vectorizing the historical multimodal parameters to obtain a historical multimodal vector; Multiplying a first model parameter of the preset recursive neural network by the historical multimodal vector through the preset recursive neural network to obtain a first multiplication result; multiplying the second model parameter of the preset recursive neural network by the memory request amount through the preset recursive neural network to obtain a second multiplication result; The predicted request amount is obtained by taking the hyperbolic tangent value of the sum of the first multiplication result and the second multiplication result through the preset recursive neural network, and adding the hyperbolic tangent value to the third model parameter of the preset recursive neural network.

[0092] Optionally, the model parameters include a first model parameter, a second model parameter and a third model parameter, and the adjustment module 30 is further used for: Comparing the actual request amount with the predicted request amount to obtain a difference value between the actual request amount and the predicted request amount; Calculating a mean square error of the preset recursive neural network based on the difference value; A first gradient of the first model parameter, a second gradient of the second model parameter, and a third gradient of the third model parameter are calculated according to the mean square error to obtain the comparison result.

[0093] Optionally, the scheduling module 40 in the interface traffic prediction device is used to: Determine a target request quantity threshold value among a plurality of preset request quantity threshold values ​​based on the predicted request quantity, and obtain a scheduling instruction corresponding to the target request quantity threshold value, wherein, when the predicted request quantity is greater than the preset expansion threshold value, the target request quantity threshold value is a preset request quantity threshold value that is adjacent to the predicted request quantity and smaller than the predicted request quantity, and when the predicted request quantity is less than the preset reduction threshold value, the target request quantity threshold value is a preset request quantity threshold value that is adjacent to the predicted request quantity and larger than the predicted request quantity; The scheduling instruction is sent to the cluster to be scheduled, so that the cluster to be scheduled adjusts the capacity of the service instances in the cluster to be scheduled based on the scheduling instruction.

[0094] Optionally, the cluster to be scheduled includes a K8s cluster, and the scheduling module 40 is further used to: When the predicted request amount is greater than a preset expansion threshold, the scheduling instruction is sent to the K8s cluster, so that the K8s cluster increases the number of service instances in the K8s cluster and / or increases the resources allocated to the service instances based on the scheduling instruction; When the predicted request amount is less than the preset shrinking threshold, the scheduling instruction is sent to the K8s cluster so that the K8s cluster reduces the number of service instances in the K8s cluster and / or reduces the resources allocated to the service instances based on the scheduling instruction.

[0095] The interface traffic prediction device provided by the present application adopts the interface traffic prediction method in the above embodiment, which can solve the technical problem of how to accurately predict the future traffic of the interface. Compared with the prior art, the beneficial effects of the interface traffic prediction device provided by the present application are the same as the beneficial effects of the interface traffic prediction method provided by the above embodiment, and other technical features in the interface traffic prediction device are the same as the features disclosed in the above embodiment method, which will not be repeated here.

[0096] The present application provides an electronic device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the interface traffic prediction method in the above-mentioned embodiment one.

[0097] Reference below Figure 7 , which shows a schematic diagram of the structure of an electronic device suitable for implementing the embodiments of the present application. The electronic devices in the embodiments of the present application may include but are not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions: tablet computers), PMPs (Portable Media Players: portable multimedia players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 7 The electronic device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0098] like Figure 7 As shown, the electronic device may include a processing device 1001 (e.g., a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM: Random Access Memory) 1004. In RAM1004, various programs and data required for the operation of the electronic device are also stored. The processing device 1001, ROM1002, and RAM1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the electronic device to communicate with other devices wirelessly or by wire to exchange data. Although the figure shows an electronic device with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be implemented or have alternatively.

[0099] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0100] The electronic device provided by the present application adopts the interface traffic prediction method in the above embodiment to solve the technical problem of how to accurately predict the future traffic of the interface. Compared with the prior art, the beneficial effects of the electronic device provided by the present application are the same as the beneficial effects of the interface traffic prediction method provided by the above embodiment, and the other technical features in the electronic device are the same as the features disclosed in the method of the previous embodiment, which will not be repeated here.

[0101] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0102] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0103] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer programs) stored thereon, and the computer-readable program instructions are used to execute the interface traffic prediction method in the above-mentioned embodiment.

[0104] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM: Random Access Memory), a read-only memory (ROM: Read Only Memory), an erasable programmable read-only memory (EPROM: Erasable Programmable Read Only Memory or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM: CD-Read Only Memory), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency: Radio Frequency), etc., or any suitable combination of the above.

[0105] The computer-readable storage medium may be included in the electronic device, or may exist independently without being installed in the electronic device.

[0106] The computer-readable storage medium carries one or more programs. When the one or more programs are executed by an electronic device, the electronic device: obtains historical traffic data of the interface, and extracts historical multimodal parameters in the historical traffic data, wherein the historical multimodal parameters at least include historical time series, historical request volume, request path, request parameters, response parameters and response time; inputs the historical multimodal parameters into a preset recursive neural network to obtain the predicted request volume of the interface at a preset time.

[0107] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0108] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0109] The modules involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.

[0110] The readable storage medium provided in the present application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned interface traffic prediction method, and can solve the technical problem of how to accurately predict the future traffic of the interface. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in the present application are the same as the beneficial effects of the interface traffic prediction method provided in the above-mentioned embodiment, and will not be repeated here.

[0111] The present application also provides a computer program product, including a computer program, which implements the steps of the above-mentioned interface traffic prediction method when executed by a processor.

[0112] The computer program product provided by this application can solve the technical problem of how to accurately predict the future traffic of an interface. Compared with the prior art, the beneficial effects of the computer program product provided by this application are the same as the beneficial effects of the interface traffic prediction method provided by the above embodiment, which will not be repeated here.

[0113] The above descriptions are only some embodiments of the present application, and are not intended to limit the patent scope of the present application. All equivalent structural changes made using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect applications in other related technical fields are included in the patent protection scope of the present application.

Claims

1. An interface traffic prediction method, characterized in that: The interface traffic prediction method comprises: Obtaining historical traffic data of the interface, and extracting historical multimodal parameters in the historical traffic data, wherein the historical multimodal parameters at least include historical time series, historical request volume, request path, request parameters, response parameters, and response time; The historical multimodal parameters are input into a preset recursive neural network to obtain the predicted request volume of the interface at a preset time.

2. The interface traffic prediction method according to claim 1, characterized in that: After the step of inputting the historical multimodal parameters into a preset recursive neural network to obtain the predicted request volume of the interface at a preset time, the method further includes: Monitoring the actual request volume of the interface at a preset time, and comparing the actual request volume with the predicted request volume to obtain a comparison result; The model parameters of the preset recursive neural network are adjusted based on the comparison result.

3. The interface traffic prediction method according to claim 1, characterized in that: The step of inputting the historical multimodal parameters into a preset recursive neural network to obtain the predicted request volume of the interface at a preset time includes: Using the last predicted request amount output by the preset recursive neural network as the memory request amount; The historical multimodal parameters and the memory request amount are input into the preset recursive neural network, and the predicted request amount is calculated in the preset recursive neural network according to the memory request amount and the historical multimodal parameters.

4. The interface traffic prediction method according to claim 3, characterized in that: The step of calculating the predicted request amount in the preset recursive neural network according to the memorized request amount and the historical multimodal parameters comprises: Vectorizing the historical multimodal parameters to obtain a historical multimodal vector; Multiplying a first model parameter of the preset recursive neural network by the historical multimodal vector through the preset recursive neural network to obtain a first multiplication result; multiplying the second model parameter of the preset recursive neural network by the memory request amount through the preset recursive neural network to obtain a second multiplication result; The predicted request amount is obtained by taking the hyperbolic tangent value of the sum of the first multiplication result and the second multiplication result through the preset recursive neural network, and adding the hyperbolic tangent value to the third model parameter of the preset recursive neural network.

5. The interface traffic prediction method according to claim 2, characterized in that: The model parameters include a first model parameter, a second model parameter and a third model parameter, and the step of comparing the actual request amount with the predicted request amount to obtain a comparison result includes: Comparing the actual request amount with the predicted request amount to obtain a difference value between the actual request amount and the predicted request amount; Calculating a mean square error of the preset recursive neural network based on the difference value; A first gradient of the first model parameter, a second gradient of the second model parameter, and a third gradient of the third model parameter are calculated according to the mean square error to obtain the comparison result.

6. The interface traffic prediction method according to any one of claims 1 to 5, characterized in that: The interface traffic prediction method also includes: Determine a target request quantity threshold value among a plurality of preset request quantity threshold values ​​based on the predicted request quantity, and obtain a scheduling instruction corresponding to the target request quantity threshold value, wherein, when the predicted request quantity is greater than the preset expansion threshold value, the target request quantity threshold value is a preset request quantity threshold value that is adjacent to the predicted request quantity and smaller than the predicted request quantity, and when the predicted request quantity is less than the preset reduction threshold value, the target request quantity threshold value is a preset request quantity threshold value that is adjacent to the predicted request quantity and larger than the predicted request quantity; The scheduling instruction is sent to the cluster to be scheduled, so that the cluster to be scheduled adjusts the capacity of the service instances in the cluster to be scheduled based on the scheduling instruction.

7. The interface traffic prediction method according to claim 6, characterized in that: The cluster to be scheduled includes a K8s cluster, and the step of sending the scheduling instruction to the cluster to be scheduled so that the cluster to be scheduled adjusts the capacity of the service instances in the cluster to be scheduled based on the scheduling instruction includes: When the predicted request amount is greater than a preset expansion threshold, the scheduling instruction is sent to the K8s cluster, so that the K8s cluster increases the number of service instances in the K8s cluster and / or increases the resources allocated to the service instances based on the scheduling instruction; When the predicted request amount is less than the preset shrinking threshold, the scheduling instruction is sent to the K8s cluster so that the K8s cluster reduces the number of service instances in the K8s cluster and / or reduces the resources allocated to the service instances based on the scheduling instruction.

8. An interface traffic prediction device, characterized in that: The interface traffic prediction device comprises: An acquisition module, used to acquire historical traffic data of an interface and extract historical multimodal parameters from the historical traffic data, wherein the historical multimodal parameters at least include historical time series, historical request volume, request path, request parameters, response parameters and response time; The output module is used to input the historical multimodal parameters into a preset recursive neural network to obtain the predicted request volume of the interface at a preset time.

9. An electronic device, characterized in that: The device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the interface traffic prediction method according to any one of claims 1 to 7.

10. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the interface traffic prediction method according to any one of claims 1 to 7 are implemented.

Images