Server processing scheme determination method and device, equipment, medium and program product
Automatically determine the server processing solution through the prediction model and decision model based on server performance parameters, the problem of inaccurate processing solutions in the prior art is solved, and the accuracy of the processing solution and the security of the server are improved.
Patent Information
- Application Number
- CN202411670540.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-20
- Publication Date
- 2025-02-28
AI Technical Summary
In the prior art, staff determine the server processing plan based on their own experience, resulting in inaccurate processing plan.
By determining the target performance parameters based on the original performance parameters of the server, input them into the fault prediction model and the performance parameter prediction model, obtaining the predicted failure rate and predicted performance parameters, and then inputting them into the processing solution decision model to obtain the target processing solution.
The processing plan is not required to be manually determined based on experience, which effectively improves the accuracy of the processing plan, reduces the difficulty for staff, and determines the processing plan before the server fails, improving the security of the server.
Smart Images
Figure CN119537111A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of computer technology, and in particular to a method, device, equipment, medium and program product for determining a server processing solution. Background Art
[0002] With its rich computing resources and large data storage capacity, servers are increasingly used in various industries. In order to ensure that the business handled by the server is not interrupted, it is necessary to monitor the server.
[0003] In the prior art, multiple parameters of the server are usually obtained, each parameter corresponds to a threshold, and an alarm is triggered when the parameter exceeds the threshold. The staff then determines a processing solution based on their own experience to process the server.
[0004] In summary, in the prior art, workers determine treatment plans based on their own experience, which results in inaccurate treatment plans. Summary of the invention
[0005] The embodiments of the present application provide a method, apparatus, device, medium and program product for determining a server processing solution, so as to solve the problem in the prior art that the processing solution is inaccurate due to the staff determining the processing solution based on their own experience.
[0006] In a first aspect, an embodiment of the present application provides a method for determining a server processing solution, including:
[0007] Determine target performance parameters based on the acquired original performance parameters of the server;
[0008] Inputting the target performance parameter into a fault prediction model to obtain a predicted failure rate, wherein the fault prediction model is a pre-trained classification model that calculates the failure rate based on the performance parameter;
[0009] Inputting the target performance parameter into a performance parameter prediction model to obtain a predicted performance parameter, wherein the performance parameter prediction model is a pre-trained time series model that calculates new performance parameters based on the input performance parameters;
[0010] The predicted failure rate and the predicted performance parameters are input into a processing solution decision model to obtain a target processing solution. The processing solution decision model is a pre-trained reinforcement learning model that determines a processing solution based on failure rate and performance parameters. The target processing solution is one of maintaining the status quo, backup, and migration.
[0011] In a second aspect, an embodiment of the present application provides a server processing solution determination device, including:
[0012] A processing module, used to determine target performance parameters according to the acquired original performance parameters of the server;
[0013] Prediction module for:
[0014] Inputting the target performance parameter into a fault prediction model to obtain a predicted failure rate, wherein the fault prediction model is a pre-trained classification model that calculates the failure rate based on the performance parameter;
[0015] Inputting the target performance parameter into a performance parameter prediction model to obtain a predicted performance parameter, wherein the performance parameter prediction model is a pre-trained time series model that calculates new performance parameters based on the input performance parameters;
[0016] The processing module is also used to input the predicted failure rate and the predicted performance parameters into a processing solution decision model to obtain a target processing solution. The processing solution decision model is a pre-trained reinforcement learning model that determines the processing solution based on the failure rate and performance parameters. The target processing solution is one of maintaining the status quo, backing up, and migrating.
[0017] In a third aspect, an embodiment of the present application provides an electronic device, including:
[0018] Processor, memory, communication interface;
[0019] The memory is used to store executable instructions of the processor;
[0020] Wherein, the processor is configured to execute the server processing solution determination method described in any one of the first aspects by executing the executable instructions.
[0021] In a fourth aspect, an embodiment of the present application provides a readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the server processing solution determination method described in any one of the first aspects is implemented.
[0022] In a fifth aspect, an embodiment of the present application provides a computer program product, including a computer program, which, when executed by a processor, is used to implement the server processing solution determination method described in any one of the first aspects.
[0023] The server processing scheme determination method, device, equipment, medium and program product provided in the embodiments of the present application, after determining the target performance parameters according to the original performance parameters of the server, the target performance parameters are respectively input into the fault prediction model and the performance parameter prediction model to obtain the predicted failure rate and the predicted performance parameters. The predicted failure rate and the predicted performance parameters are then input into the processing scheme decision model to obtain the target processing scheme. This scheme first determines the predicted failure rate and the predicted performance parameters, and then uses the processing scheme decision model to process and obtain the target processing scheme, without manually determining the processing scheme based on experience, which effectively improves the accuracy of the processing scheme. BRIEF DESCRIPTION OF THE DRAWINGS
[0024] 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.
[0025] Figure 1 A schematic diagram of an application scenario of the method for determining a server processing solution provided in this application;
[0026] Figure 2a A flowchart of a method for determining a server processing solution provided in the present application;
[0027] Figure 2b A schematic diagram of the process for determining the target treatment solution provided for this application;
[0028] Figure 3a A flowchart of a second embodiment of a method for determining a server processing solution provided in the present application;
[0029] Figure 3b A schematic diagram of the training process of the processing solution decision model provided for this application;
[0030] Figure 4 A flowchart of a third embodiment of the method for determining a server processing solution provided in the present application;
[0031] Figure 5 A flowchart of a fourth embodiment of a method for determining a server processing solution provided in the present application;
[0032] Figure 6 A schematic diagram of the structure of an embodiment of a device for determining a server processing solution provided by the present application;
[0033] Figure 7 A schematic diagram of the structure of an electronic device provided in this application.
[0034] The above drawings have shown clear embodiments of the present application, which will be described in more detail later. These drawings and text descriptions are not intended to limit the scope of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION
[0035] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.
[0036] The terms "first", "second", "third", "fourth", etc. (if any) 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 sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. 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 that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.
[0037] A server is a device that provides computing power and runs software applications. With its abundant computing resources and large data storage capacity, it is increasingly used in various industries. In order to ensure that the business handled by the server is not interrupted, the server needs to be monitored.
[0038] In the prior art, multiple parameters of the server are usually obtained, each parameter corresponds to a threshold, and an alarm is triggered when the parameter exceeds the threshold. The staff then determines which parameters exceed the threshold and which processing solution should be adopted to process the server process based on their own experience. Since the staff determines the processing solution based on their own experience, the processing solution may be inaccurate.
[0039] In view of the problems existing in the prior art, the inventors found in the process of studying the method for determining the server processing solution that in order to make the obtained processing solution more accurate and to determine the processing solution without the experience of the staff, the predicted failure rate and predicted performance parameters can be determined based on the performance parameters of the server, and then the predicted failure rate and predicted performance parameters can be processed using the trained reinforcement learning model to obtain a processing solution, which is one of maintaining the status quo, backup, and migration. Based on the above invention concept, the server processing solution determination solution in this application is designed.
[0040] The executor of the method for determining the server processing solution in the present application may be a computer, or a server, a terminal device, etc. The present application does not limit it and the following description will be made using a terminal device as an example.
[0041] For example, Figure 1 A schematic diagram of an application scenario of the method for determining a server processing solution provided in this application, such as Figure 1 As shown, the application scenario may include: a server 11 and a terminal device 12 .
[0042] exist Figure 1In the application scenario shown, the terminal device 12 monitors the server 11 to determine a processing solution for the server 11 .
[0043] During operation, the server 11 sends its original performance parameters to the terminal device 12 in real time. After receiving the original performance parameters, the terminal device 12 calculates the target performance parameters according to the original performance parameters.
[0044] The terminal device 12 inputs the target performance parameter into the fault prediction model to obtain the predicted fault rate. The target performance parameter is input into the performance parameter prediction model to obtain the predicted performance parameter.
[0045] The terminal device 12 then inputs the predicted failure rate and the predicted performance parameters into the processing solution decision model to obtain a target processing solution, which is one of maintaining the status quo, backing up, and migrating.
[0046] The terminal device 12 can display the measured failure rate, predicted performance parameters and target processing solutions, and the staff can process the server 11 according to the target processing solutions.
[0047] It should be noted that Figure 1 This is only a schematic diagram of an application scenario provided by the embodiment of the present application. Figure 1 The actual form of the various devices included in the Figure 1 The interaction mode between devices is limited, and in the specific application of the solution, it can be set according to actual needs.
[0048] The technical solution of the present application is described in detail below through specific embodiments. It should be noted that the following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.
[0049] Figure 2a This is a flow chart of the first embodiment of the method for determining a server processing solution provided by the present application. The present embodiment of the present application describes the situation in which the terminal device obtains the predicted failure rate and predicted performance parameters based on the fault prediction model and the performance parameter prediction model, and then obtains the target processing solution in combination with the processing solution decision model. The method in this embodiment can be implemented by software, hardware, or a combination of software and hardware. Figure 2a As shown, the server processing solution determination method specifically includes the following steps:
[0050] S201: Determine target performance parameters according to the acquired original performance parameters of the server.
[0051] In this step, in order to predict whether the server fails and the treatment plan to be taken, the terminal device needs to obtain the original performance parameters of the server and then determine the target performance parameters based on the original performance parameters.
[0052] In order to reduce the amount of calculation of the performance parameter prediction model and the fault prediction model and improve the calculation efficiency, it is necessary to reduce the number of performance parameters, and also ensure that the performance parameters can calculate the failure rate and predict future performance parameters. Therefore, the target performance parameters are determined based on the original performance parameters to reduce the number of performance parameters. And because the target performance parameters cannot be directly obtained, the target performance parameters need to be calculated from the original performance parameters.
[0053] Specifically, the original performance parameters include central processing unit (CPU) parameters, memory parameters, disk parameters and network parameters, and the target performance parameters include CPU utilization, CPU load, memory usage, disk read rate, disk write rate and network throughput.
[0054] It should be noted that the preset historical moment is the moment before the current moment, which can be 1 minute, 5 minutes, or 20 minutes before the current moment. The embodiment of the present application does not limit the preset historical moment, and it can be determined according to actual conditions.
[0055] Based on the CPU parameters, determine the CPU utilization and CPU load.
[0056] The CPU parameters include the cumulative time of running user space between the preset historical moment and the current moment, the cumulative time of running user space with adjusted priority between the preset historical moment and the current moment, the cumulative time of running kernel between the preset historical moment and the current moment, the cumulative idle time between the preset historical moment and the current moment, the cumulative time of waiting for input / output requests between the preset historical moment and the current moment, the cumulative time of processing hardware interrupts between the preset historical moment and the current moment, the cumulative time of processing software interrupts between the preset historical moment and the current moment, and the CPU load at the current moment;
[0057] The CPU utilization is calculated using the formula U=1-T1 / T2, where T1=T3-T4, T3=a1+b1+c1+d1+e1+f1+g1, T4=a2+b2+c2+d2+e2+f2+g2, T2=d1-d2, U represents the CPU utilization, a1 represents the current cumulative user space running time, b1 represents the current cumulative user space running priority adjustment time, c1 represents the current cumulative kernel running time, d1 represents the current idle cumulative time, e1 represents the current cumulative waiting input / output request time. The timing length is: f1 represents the cumulative time of processing hardware interrupts at the current moment, g1 represents the cumulative time of processing software interrupts at the current moment, a2 represents the cumulative time of running user space at a preset historical moment, b2 represents the cumulative time of running user space with adjusted priority at a preset historical moment, c2 represents the cumulative time of running kernel at a preset historical moment, d2 represents the cumulative time of idle time at a preset historical moment, e2 represents the cumulative time of waiting for input / output requests at a preset historical moment, f2 represents the cumulative time of processing hardware interrupts at a preset historical moment, and g2 represents the cumulative time of processing software interrupts at a preset historical moment.
[0058] The current CPU load is used as the CPU load in the target performance parameter.
[0059] It should be noted that if there are multiple CPU loads at the current moment, representing the average CPU loads in different durations, one can be randomly selected as the CPU load in the target performance parameter, or the average CPU load in a preset duration can be selected as the CPU load in the target performance parameter. The preset duration can be 1 minute, 5 minutes, 15 minutes, etc. The embodiment of the present application does not limit the preset duration, and it can be determined according to actual conditions.
[0060] Based on the memory parameters, determine the memory usage.
[0061] The memory parameters include the current memory usage capacity and the total memory capacity. The ratio of the current memory usage capacity to the current memory total capacity is taken as the memory usage rate.
[0062] Based on the disk parameters, determine the disk read rate and disk write rate.
[0063] The disk parameters include the cumulative number of sector reads between the preset historical moment and the current moment, and the cumulative number of sector writes between the preset historical moment and the current moment.
[0064] The disk read rate is calculated using the formula S1=r1-r2 / t1-t2, where S1 represents the disk read rate, r1 represents the cumulative number of sector reads at the current moment, r2 represents the cumulative number of sector reads at a preset historical moment, t1 represents the current moment, t2 represents the preset historical moment, and t1-t2 represents the duration between the preset historical moment and the current moment.
[0065] The disk read rate is calculated using the formula S2=r3-r4 / t1-t2, where S2 represents the disk write rate, r3 represents the cumulative number of sector writes at the current moment, r4 represents the cumulative number of sector writes at a preset historical moment, t1 represents the current moment, t2 represents the preset historical moment, and t1-t2 represents the duration between the preset historical moment and the current moment.
[0066] Determine the network throughput based on the network parameters.
[0067] The network parameters include the byte sending rate and the byte receiving rate at the current moment. The sum of the byte sending rate and the byte receiving rate at the current moment is taken as the network throughput.
[0068] It should be noted that the way in which the terminal device obtains the original performance parameters can be: the server sends its own original performance parameters to the terminal device in real time or periodically, and the terminal device can obtain the original performance parameters. It can also be: the terminal device sends a parameter acquisition request to the server, and the server sends its own original performance parameters to the terminal device, and the terminal device can obtain the original performance parameters. It can also be: the staff sends the original performance parameters of the server to the terminal device, and the terminal device can obtain the original performance parameters. The embodiment of the present application does not limit the way in which the terminal device obtains the original performance parameters, which can be determined according to actual conditions.
[0069] S202: Input the target performance parameter into the fault prediction model to obtain the predicted failure rate.
[0070] In this step, after the terminal device determines the target performance parameters, in order to improve the accuracy of the subsequent processing solution, it is necessary to determine the predicted failure rate, input the target performance parameters into the fault prediction model, and obtain the predicted failure rate.
[0071] The fault prediction model is a pre-trained classification model that calculates the failure rate based on performance parameters.
[0072] It should be noted that the fault prediction model can be a support vector machine (SVM) model, a random forest model, a K-nearest neighbor algorithm, etc. The embodiment of the present application does not limit the fault prediction model and can be determined according to actual conditions.
[0073] Using classification models to process target performance parameters can improve the accuracy of predicting failure rates.
[0074] S203: Input the target performance parameter into the performance parameter prediction model to obtain the predicted performance parameter.
[0075] In the step, after the terminal device determines the target performance parameters, in order to improve the accuracy of the processing solution determined subsequently, it is necessary to determine the predicted performance parameters, input the target performance parameters into the performance parameter prediction model, and obtain the predicted performance parameters.
[0076] The performance parameter prediction model is a pre-trained time series model that calculates new performance parameters based on input performance parameters.
[0077] It should be noted that the predicted performance parameters are the performance parameters of the server in the future, including CPU utilization, CPU load, memory usage, disk read rate, disk write rate, and network throughput.
[0078] It should be noted that the performance parameter prediction model can be a long short-term memory (LSTM) network model, a gated recurrent unit (GRU) model, a vector autoregression (VAR) model, etc. The embodiment of the present application does not limit the performance parameter prediction model and can be determined according to actual conditions.
[0079] Since the performance parameters of the server are time series data, the target performance parameters are processed using a time series model to improve the accuracy of the predicted performance parameters.
[0080] It should be noted that the execution order of step S202 and step S203 can be: first execute step S202, then execute step S203; or: first execute step S203, then execute step S203; or: execute step S202 and step S203 simultaneously. The embodiment of the present application does not limit the execution order of step S202 and step S203, which can be determined according to actual conditions.
[0081] S204: Input the predicted failure rate and the predicted performance parameters into the processing solution decision model to obtain a target processing solution.
[0082] In this step, after the terminal device determines the predicted failure rate and the predicted performance parameters, the predicted failure rate and the predicted performance parameters are input into the processing solution decision model to obtain the target processing solution.
[0083] The processing solution decision model is a pre-trained reinforcement learning model that determines the processing solution based on the failure rate and performance parameters, and the target processing solution is one of maintaining the status quo, backing up, and migrating.
[0084] If the target solution is to maintain the current status, it means that the server is not faulty. If the target solution is to backup or migrate, it means that the server is faulty.
[0085] For example, Figure 2b A flow chart of the target treatment scheme provided for this application is as follows: Figure 2b As shown, the terminal device determines the target performance parameters based on the original performance parameters of the server obtained; then inputs the target performance parameters into the fault prediction model to obtain the predicted failure rate; inputs the target performance parameters into the performance parameter prediction model to obtain the predicted performance parameters; then inputs the predicted failure rate and the predicted performance parameters into the processing solution decision model to obtain the target processing solution.
[0086] The terminal device can then output the target processing solution, predicted failure rate, and predicted performance parameters so that the staff can use the target processing solution to process the server before the failure occurs, thus improving the security of the data in the server. The staff can also analyze the server based on the predicted failure rate and predicted performance parameters so as to repair the server, thus improving the security of the server.
[0087] The server processing solution determination method provided in this embodiment determines the target performance parameters according to the original performance parameters of the server, and then inputs the target performance parameters into the fault prediction model and the performance parameter prediction model respectively to obtain the predicted failure rate and the predicted performance parameters. The predicted failure rate and the predicted performance parameters are then input into the processing solution decision model to obtain the target processing solution. This solution first determines the predicted failure rate and the predicted performance parameters, and then uses the processing solution decision model to process and obtain the target processing solution. There is no need to manually determine the processing solution based on experience, which effectively improves the accuracy of the processing solution.
[0088] In addition, this solution does not require the staff to understand the relevant knowledge of the performance parameters of the server. The target processing solution can be obtained by processing the performance parameters through the terminal device. The staff can process the server according to the target processing solution, which reduces the difficulty of obtaining the target processing solution. In addition, this solution determines the target processing solution by predicting the failure rate and the predicted performance parameters, so that the target processing solution is determined before the server fails, which improves the security of the server. Compared with the prior art of regularly backing up the server, this solution backs up when the target processing solution is a backup, saving server resources.
[0089] Figure 3aThis is a flow chart of the second embodiment of the method for determining a server processing solution provided by the present application. Based on the above embodiment, the present embodiment of the present application illustrates the situation where the terminal device trains the initial reinforcement learning model to obtain a processing solution decision model. Figure 3a As shown, the server processing solution determination method specifically includes the following steps:
[0090] S301: Obtain a training data sequence.
[0091] In this step, in order to train the treatment solution decision model, it is necessary to obtain a training data sequence.
[0092] Each training data in the training data sequence includes a failure rate and performance parameters, and the performance parameters include CPU utilization, CPU load, memory usage, disk read rate, disk write rate, and network throughput.
[0093] S302: taking the first training data in the training data as the target training data.
[0094] In this step, since the training data in the training data sequence is a sequence obtained after the staff sorts the training data, it is necessary to select the training data in sequence for training according to the order of the training data in the training data sequence, which can improve the accuracy of the processing solution decision model, so the first training data in the training data is first used as the target training data.
[0095] S303: Input the target training data into the initial reinforcement learning model to obtain a training processing solution.
[0096] In this step, after the terminal device determines the target training data, it trains the initial reinforcement learning model according to the target training data and the training data sequence to obtain a processing solution decision model. First, the target training data is input into the initial reinforcement learning model to obtain a training processing solution, which is one of maintaining the status quo, backup, and migration.
[0097] S304: Determine a training feedback value according to the preset feedback function corresponding to the target training data and the training processing scheme.
[0098] In this step, after the terminal device determines the training processing scheme, it determines the training feedback value according to the target training data and the preset feedback function corresponding to the training processing scheme.
[0099] Specifically, if the training processing scheme is to maintain the status quo, the preset feedback function corresponding to the training processing scheme is Where R1 represents the training feedback value, x represents the sum of the failure rate, CPU utilization, CPU load, memory usage, disk read rate, disk write rate and network throughput in the target training data, and b1, a, and m1 are preset values;
[0100] If the training processing scheme is backup, the preset feedback function corresponding to the training processing scheme is Where R2 represents the training feedback value, x represents the sum of the failure rate, CPU utilization, CPU load, memory usage, disk read rate, disk write rate and network throughput in the target training data, b2, a, and m2 are preset values, and b2 is less than b1;
[0101] If the training solution is migration, the preset feedback function corresponding to the training solution is Where R3 represents the training feedback value, x represents the sum of the failure rate, CPU utilization, CPU load, memory usage, disk read rate, disk write rate and network throughput in the target training data, b3, a, and m3 are preset values, and b3 is less than b1.
[0102] It should be noted that b1, a, m1, b2, m2, b3, and m3 can be 0.1, 1, 5, 10, 50, 100, 200, 500, etc. The embodiments of the present application do not limit b1, a, m1, b2, m2, b3, and m3, and can be determined according to actual conditions.
[0103] S305: Update the initial reinforcement learning model according to the target training data, the next training data of the target training data in the training data sequence, the training feedback value, and the training processing scheme to obtain a trained reinforcement learning model.
[0104] In this step, after the terminal device determines the training feedback value, it updates the initial reinforcement learning model according to the target training data, the next training data of the target training data in the training data sequence, the training feedback value, and the training processing plan to obtain the trained reinforcement learning model.
[0105] S306: Update the number of training times.
[0106] In this step, after the terminal device obtains the trained reinforcement learning model, it updates the number of training times, that is, increases the number of training times by one.
[0107] S307: Determine whether the updated training times are equal to the preset times. If the updated training times are less than the preset times, execute step S308; if the updated training times are equal to the preset times, execute step S309.
[0108] In this step, after the terminal device updates the number of training times, in order to determine whether to continue model training, it is necessary to determine whether the updated number of training times is equal to the preset number.
[0109] It should be noted that the preset number of times is less than the number of training data in the training data sequence, and the preset number of times may be 5000, 1000, 50000, etc. The embodiment of the present application does not limit the preset number of times, and it may be determined according to actual conditions.
[0110] S308: Use the next training data of the target training data in the training data sequence as the new target training data, use the trained reinforcement learning model as the new initial reinforcement learning model, and execute step S303.
[0111] In this step, if the terminal device determines that the updated training times are less than the preset times, it means that the model training needs to continue, and the next training data of the target training data in the training data sequence is used as the new target training data, and the trained reinforcement learning model is used as the new initial reinforcement learning model, and then step S303 is executed.
[0112] That is, the new target training data is input into the initial reinforcement learning model to obtain the training processing plan, the training feedback value is re-determined, the new initial reinforcement learning model is updated, the trained reinforcement learning model is obtained, and the number of training times is updated. If the updated number of training times is less than the preset number, the next training data of the target training data in the training data sequence is used as the new target training data, and the trained reinforcement learning model is used as the new initial reinforcement learning model. This process is repeated until the updated number of training times is equal to the preset number, and the trained reinforcement learning model is used as the processing plan decision model.
[0113] S309: Using the trained reinforcement learning model as a processing solution decision model.
[0114] In this step, if the terminal device determines that the updated training times are equal to the preset times, it means that there is no need to continue model training, and the trained reinforcement learning model is used as the processing solution decision model.
[0115] For example, Figure 3b A schematic diagram of the training process of the processing solution decision model provided for this application, such as Figure 3b As shown in the figure, the initial reinforcement learning model can be used as an intelligent agent, and the terminal device provides a training environment. The training environment inputs the selected target training data as the state to the intelligent agent. The state can be represented by a 7-tuple, and the data in the 7-tuple are failure rate, CPU utilization, CPU load, memory usage, disk read rate, disk write rate and network throughput.
[0116] The agent processes the state to obtain a training solution, that is, to determine the action corresponding to the state from the action space. The training solution is also called an action. The action space is represented by a 3-tuple, and the data in the 3-tuple is to maintain the status quo, backup, and migration.
[0117] The training environment calculates feedback values based on the state and action, and then inputs the feedback values and the next state into the agent. The next state is the next training data of the target training data in the training data sequence.
[0118] The intelligent agent is updated according to the state, feedback value, next state, and action to obtain a new intelligent agent. The new intelligent agent processes the next state to obtain a new training processing plan, that is, to determine the action corresponding to the next state from the action space.
[0119] Repeat the above process until the number of training times is equal to the preset number of times, and obtain the processing solution decision model.
[0120] The server processing solution determination method provided in this embodiment can improve the accuracy of the training feedback value and further improve the accuracy of the processing solution decision model by determining the training feedback value according to the preset feedback function corresponding to the target training data and the training processing solution.
[0121] Figure 4 This is a flow chart of the third embodiment of the method for determining the server processing solution provided by the present application. Based on the above embodiment, the present embodiment of the present application illustrates the situation where the terminal device trains the initial classification model to obtain the fault prediction model. Figure 4 As shown, the server processing solution determination method specifically includes the following steps:
[0122] S401: Acquire a plurality of first data.
[0123] In this step, in order to train a fault prediction model, the terminal device needs to first obtain a plurality of first data.
[0124] The data included in each first data is of the same data type as the data included in the target performance parameter.
[0125] S402: Select one data from a plurality of first data as first target data.
[0126] S403: Input the first target data into the initial classification model to obtain a training failure rate.
[0127] In the above steps, after the terminal device obtains the first target data, it selects one data from multiple first data as the first target data, and then inputs the first target data into the initial classification model to obtain the training failure rate.
[0128] S404: updating the initial classification model according to the training failure rate to obtain a trained classification model.
[0129] In this step, after the terminal device obtains the training failure rate, it updates the initial classification model according to the training failure rate to obtain the trained classification model. That is, the loss value is calculated according to the training failure rate and the actual failure rate corresponding to the first target data, and the loss value is used to update the initial classification model to obtain the trained classification model.
[0130] S405: Update the number of training times.
[0131] In this step, the terminal device obtains the trained classification model and updates the number of training times, that is, increases the number of training times by one.
[0132] S406: Determine whether the updated training times are equal to the first preset training times. If the updated training times are less than the first preset training times, execute step S407; if the updated training times are equal to the first preset training times, execute step S408.
[0133] In this step, after the terminal device updates the number of training times, in order to determine whether to continue model training, it is necessary to determine whether the updated number of training times is equal to the first preset number of training times.
[0134] It should be noted that the first preset number of training times may be 5000, 1000, 50000, etc. The embodiment of the present application does not limit the first preset number of training times, which may be determined according to actual conditions.
[0135] S407: Use the trained classification model as a new initial classification model and execute step S402.
[0136] In this step, if the terminal device determines that the updated training times are less than the first preset training times, it means that the model training needs to be continued, and the trained classification model is used as a new initial classification model, and then step S402 is executed.
[0137] That is, one data is reselected from multiple first data as the new first target data, the new first target data is input into the new initial classification model to obtain a new training failure rate, and the new initial classification model is updated according to the new training failure rate to obtain a trained classification model, and the number of training times is updated. If the updated number of training times is less than the first preset number of training times, the trained classification model is used as the new initial classification model, and this process is repeated until the updated number of training times is equal to the first preset number of training times, and the trained classification model is used as the fault prediction model.
[0138] S408: Using the trained classification model as a fault prediction model.
[0139] In this step, if the terminal device determines that the updated training times are equal to the first preset training times, it means that there is no need to continue model training, and the trained classification model is used as the fault prediction model.
[0140] The server processing solution determination method provided in this embodiment trains the initial classification model through first data of the same data type as the target performance parameter to obtain a fault prediction model, thereby improving the accuracy of the fault prediction model.
[0141] Figure 5 This is a flow chart of the fourth embodiment of the method for determining the server processing solution provided by the present application. Based on the above embodiment, the present embodiment of the present application illustrates the situation where the terminal device trains the initial time series model to obtain the performance parameter prediction model. Figure 5 As shown, the server processing solution determination method specifically includes the following steps:
[0142] S501: Acquire a plurality of second data.
[0143] In this step, in order to train a performance parameter prediction model, the terminal device needs to first obtain a plurality of second data.
[0144] The data included in each second data is of the same data type as the data included in the target performance parameter.
[0145] S502: Select one data from a plurality of second data as second target data.
[0146] S503: Input the second target data into the initial time series model to obtain training performance parameters.
[0147] In the above steps, after the terminal device obtains the second target data, it selects one data from multiple second data as the second target data, and then inputs the second target data into the initial time series model to obtain the training performance parameters.
[0148] S504: Update the initial time series model according to the training performance parameters to obtain a trained time series model.
[0149] In this step, after the terminal device obtains the training performance parameters, it updates the initial time series model according to the training performance parameters to obtain the trained time series model. That is, the loss value is calculated according to the training performance parameters and the real performance parameters corresponding to the second target data, and the initial time series model is updated using the loss value to obtain the trained time series model.
[0150] S505: Update the number of training times.
[0151] In this step, the terminal device obtains the trained classification model and updates the number of training times, that is, increases the number of training times by one.
[0152] S506: Determine whether the updated training number is equal to the second preset training number. If the updated training number is less than the second preset training number, execute step S507; if the updated training number is equal to the second preset training number, execute step S508.
[0153] In this step, after the terminal device updates the number of training times, in order to determine whether to continue model training, it is necessary to determine whether the updated number of training times is equal to the second preset number of training times.
[0154] It should be noted that the second preset number of training times may be 5000, 1000, 50000, etc. The embodiment of the present application does not limit the second preset number of training times, which may be determined according to actual conditions.
[0155] S507: Use the trained time series model as a new initial time series model and execute step S502.
[0156] In this step, if the terminal device determines that the updated training times are less than the second preset training times, it means that the model training needs to be continued, and the trained time series model is used as the new initial time series model, and then step S502 is executed.
[0157] That is, one data is reselected from multiple second data as the new second target data, the new second target data is input into the new initial time series model to obtain new training performance parameters, the new initial time series model is updated according to the new training performance parameters to obtain the trained time series model, the training times are updated, if the updated training times are less than the second preset training times, the trained time series model is used as the new initial time series model, and this process is repeated until the updated training times are equal to the second preset training times, and the trained time series model is used as the performance parameter prediction model.
[0158] S508: Using the trained time series model as a performance parameter prediction model.
[0159] In this step, if the terminal device determines that the updated training times are equal to the second preset training times, it means that there is no need to continue model training, and the trained classification model is used as the fault prediction model.
[0160] The server processing solution determination method provided in this embodiment trains the initial time series model through second data of the same data type as the target performance parameter to obtain a performance parameter prediction model, thereby improving the accuracy of the performance parameter prediction model.
[0161] The following is an embodiment of the device of the present application, which can be used to execute the embodiment of the method of the present application. For details not disclosed in the embodiment of the device of the present application, please refer to the embodiment of the method of the present application.
[0162] Figure 6 This is a schematic diagram of the structure of an embodiment of the server processing solution determination device provided in this application. Figure 6 As shown, the server processing solution determining device 60 includes:
[0163] The processing module 61 is used to determine the target performance parameter according to the acquired original performance parameter of the server;
[0164] The prediction module 62 is used to:
[0165] Inputting the target performance parameter into a fault prediction model to obtain a predicted failure rate, wherein the fault prediction model is a pre-trained classification model that calculates the failure rate based on the performance parameter;
[0166] Inputting the target performance parameter into a performance parameter prediction model to obtain a predicted performance parameter, wherein the performance parameter prediction model is a pre-trained time series model that calculates new performance parameters based on the input performance parameters;
[0167] The processing module 63 is also used to input the predicted failure rate and the predicted performance parameters into a processing solution decision model to obtain a target processing solution. The processing solution decision model is a pre-trained reinforcement learning model that determines the processing solution based on the failure rate and performance parameters. The target processing solution is one of maintaining the status quo, backing up, and migrating.
[0168] Furthermore, the original performance parameters include central processing unit (CPU) parameters, memory parameters, disk parameters and network parameters, and the target performance parameters include CPU utilization, CPU load, memory usage, disk read rate, disk write rate and network throughput;
[0169] When the processing module 61 is used to determine the target performance parameter according to the acquired original performance parameter of the server, it is specifically used to:
[0170] Determining the CPU utilization and the CPU load according to the CPU parameters;
[0171] Determining the memory usage rate according to the memory parameter;
[0172] Determining the disk read rate and the disk write rate according to the disk parameters;
[0173] The network throughput is determined according to the network parameters.
[0174] Further, the CPU parameters include the cumulative time of running user space between the preset historical moment and the current moment, the cumulative time of running user space with adjusted priority between the preset historical moment and the current moment, the cumulative time of running kernel between the preset historical moment and the current moment, the cumulative idle time between the preset historical moment and the current moment, the cumulative time of waiting for input / output requests between the preset historical moment and the current moment, the cumulative time of processing hardware interrupts between the preset historical moment and the current moment, the cumulative time of processing software interrupts between the preset historical moment and the current moment, and the CPU load at the current moment;
[0175] The memory parameters include the used memory capacity and the total memory capacity at the current moment;
[0176] The disk parameters include the cumulative number of sector reads between the preset historical moment and the current moment, and the cumulative number of sector writes between the preset historical moment and the current moment;
[0177] The network parameters include the byte sending rate and the byte receiving rate at the current moment.
[0178] Furthermore, the fault prediction model is a SVM model;
[0179] The performance parameter prediction model is an LSTM network model.
[0180] Furthermore, the processing module 61 is also used for:
[0181] Acquire a training data sequence, each training data in the training data sequence includes a failure rate and a performance parameter, and the performance parameter includes a CPU utilization rate, a CPU load, a memory usage rate, a disk read rate, a disk write rate, and a network throughput;
[0182] Using the first training data in the training data as target training data;
[0183] The initial reinforcement learning model is trained according to the target training data and the training data sequence to obtain the processing solution decision model.
[0184] Furthermore, when the processing module 61 is used to train the initial reinforcement learning model according to the target training data and the training data sequence to obtain the processing solution decision model, it is specifically used to:
[0185] Input the target training data into the initial reinforcement learning model to obtain a training processing scheme, wherein the training processing scheme is one of maintaining the status quo, backing up, and migrating;
[0186] Determining a training feedback value according to the target training data and a preset feedback function corresponding to the training processing scheme;
[0187] updating the initial reinforcement learning model according to the target training data, the next training data of the target training data in the training data sequence, the training feedback value, and the training processing scheme to obtain a trained reinforcement learning model;
[0188] Update the number of training times;
[0189] Determine whether the updated training times are equal to the preset times;
[0190] If the updated number of training times is less than the preset number of times, the next training data of the target training data in the training data sequence is used as the new target training data, and the trained reinforcement learning model is used as the new initial reinforcement learning model. The above steps are repeated until the updated number of training times is equal to the preset number of times, and the trained reinforcement learning model is used as the processing solution decision model.
[0191] Furthermore, if the training processing scheme is to maintain the status quo, the preset feedback function corresponding to the training processing scheme is Wherein R1 represents the training feedback value, x represents the sum of the failure rate, CPU utilization, CPU load, memory usage, disk read rate, disk write rate and network throughput in the target training data, and b1, a, and m1 are preset values;
[0192] If the training processing scheme is a backup, the preset feedback function corresponding to the training processing scheme is Wherein R2 represents the training feedback value, x represents the sum of the failure rate, CPU utilization, CPU load, memory usage, disk read rate, disk write rate and network throughput in the target training data, b2, a, m2 are preset values, and b2 is less than b1;
[0193] If the training processing scheme is migration, the preset feedback function corresponding to the training processing scheme is Wherein R3 represents the training feedback value, x represents the sum of the failure rate, CPU utilization, CPU load, memory usage, disk read rate, disk write rate and network throughput in the target training data, b3, a, m3 are preset values, and b3 is less than b1.
[0194] The server processing solution determination device provided in this embodiment is used to execute the technical solution in any of the aforementioned method embodiments. Its implementation principle and technical effects are similar and will not be repeated here.
[0195] Figure 7This is a schematic diagram of the structure of an electronic device provided in this application. Figure 7 As shown, the electronic device 70 includes:
[0196] Processor 71, memory 72, and communication interface 73;
[0197] The memory 72 is used to store executable instructions of the processor 71;
[0198] The processor 71 is configured to execute the technical solution in any of the aforementioned method embodiments by executing the executable instructions.
[0199] Optionally, the memory 72 can be independent or integrated with the processor 71.
[0200] Optionally, when the memory 72 is a device independent of the processor 71, the electronic device 70 may further include:
[0201] The bus 74 , the memory 72 and the communication interface 73 are connected to the processor 71 via the bus 74 and communicate with each other. The communication interface 73 is used to communicate with other devices.
[0202] Optionally, the communication interface 73 may be implemented by a transceiver. The communication interface is used to implement communication between the database access device and other devices (such as a client, a read-write library, and a read-only library). The memory may include a random access memory (RAM) and may also include a non-volatile memory (non-volatile memory), such as at least one disk storage.
[0203] The bus 74 may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.
[0204] The above-mentioned processor can be a general-purpose processor, including a central processing unit CPU, a network processor (NP), etc.; it can also be a digital signal processor DSP, an application-specific integrated circuit ASIC, a field programmable gate array FPGA or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.
[0205] The electronic device is used to execute the technical solution in any of the aforementioned method embodiments, and its implementation principle and technical effect are similar and will not be repeated here.
[0206] An embodiment of the present application also provides a readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the technical solution provided by any of the aforementioned method embodiments is implemented.
[0207] The above-mentioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk. The readable storage medium can be any available medium that can be accessed by a general or special-purpose computer.
[0208] An exemplary readable storage medium is coupled to a processor so that the processor can read information from the readable storage medium and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can be located in an application specific integrated circuit (Application Specific Integrated Circuits, referred to as: ASIC). Of course, the processor and the readable storage medium can also exist in the device as discrete components.
[0209] The division of units is only a logical function division, and there may be other divisions in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interface, device or unit, which can be electrical, mechanical or other forms.
[0210] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0211] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.
[0212] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.
[0213] An embodiment of the present application also provides a computer program product, including a computer program, which is used to implement the technical solution provided by any of the aforementioned method embodiments when executed by a processor.
[0214] Those skilled in the art can understand that all or part of the steps of implementing the above-mentioned method embodiments can be completed by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, the steps of the above-mentioned method embodiments are executed; and the aforementioned storage medium includes: ROM, RAM, disk or optical disk and other media that can store program codes.
[0215] Finally, it should be noted that those skilled in the art will readily conceive of other embodiments of the present invention after considering the specification and practicing the invention disclosed herein. The present invention is intended to cover any variations, uses or adaptations of the present invention, which follow the general principles of the present invention and include common knowledge or customary technical means in the art not disclosed by the present invention, are not limited to the precise structure described above and shown in the drawings, and may be modified and changed in various ways without departing from the scope thereof. The scope of the present invention is limited only by the appended claims.
Claims
1. A method for determining a server processing solution, characterized in that: include: Determine target performance parameters based on the acquired original performance parameters of the server; Inputting the target performance parameter into a fault prediction model to obtain a predicted failure rate, wherein the fault prediction model is a pre-trained classification model that calculates the failure rate based on the performance parameter; Inputting the target performance parameter into a performance parameter prediction model to obtain a predicted performance parameter, wherein the performance parameter prediction model is a pre-trained time series model that calculates new performance parameters based on the input performance parameters; The predicted failure rate and the predicted performance parameters are input into a processing solution decision model to obtain a target processing solution. The processing solution decision model is a pre-trained reinforcement learning model that determines a processing solution based on failure rate and performance parameters. The target processing solution is one of maintaining the status quo, backup, and migration.
2. The method according to claim 1, characterized in that The original performance parameters include central processing unit (CPU) parameters, memory parameters, disk parameters and network parameters, and the target performance parameters include CPU utilization, CPU load, memory usage, disk read rate, disk write rate and network throughput; The step of determining the target performance parameter according to the acquired original performance parameter of the server includes: Determining the CPU utilization and the CPU load according to the CPU parameters; Determining the memory usage rate according to the memory parameter; Determining the disk read rate and the disk write rate according to the disk parameters; The network throughput is determined according to the network parameters.
3. The method according to claim 2, characterized in that The CPU parameters include the cumulative time of running user space between the preset historical moment and the current moment, the cumulative time of running user space with adjusted priority between the preset historical moment and the current moment, the cumulative time of running kernel between the preset historical moment and the current moment, the cumulative idle time between the preset historical moment and the current moment, the cumulative time of waiting for input / output requests between the preset historical moment and the current moment, the cumulative time of processing hardware interrupts between the preset historical moment and the current moment, the cumulative time of processing software interrupts between the preset historical moment and the current moment, and the CPU load at the current moment; The memory parameters include the used memory capacity and the total memory capacity at the current moment; The disk parameters include the cumulative number of sector reads between the preset historical moment and the current moment, and the cumulative number of sector writes between the preset historical moment and the current moment; The network parameters include the byte sending rate and the byte receiving rate at the current moment.
4. The method according to claim 1, characterized in that: The fault prediction model is a support vector machine (SVM) model; The performance parameter prediction model is a long short-term memory (LSTM) network model.
5. The method according to claim 1, characterized in that Before determining the target performance parameters according to the acquired original performance parameters of the server, the method further includes: Acquire a training data sequence, each training data in the training data sequence includes a failure rate and a performance parameter, and the performance parameter includes a CPU utilization rate, a CPU load, a memory usage rate, a disk read rate, a disk write rate, and a network throughput; Using the first training data in the training data as target training data; The initial reinforcement learning model is trained according to the target training data and the training data sequence to obtain the processing solution decision model.
6. The method according to claim 5, characterized in that The initial reinforcement learning model is trained according to the target training data and the training data sequence to obtain the processing solution decision model, including: Input the target training data into the initial reinforcement learning model to obtain a training processing scheme, wherein the training processing scheme is one of maintaining the status quo, backing up, and migrating; Determining a training feedback value according to the target training data and a preset feedback function corresponding to the training processing scheme; According to the target training data, the next training data of the target training data in the training data sequence, the training feedback value, and the training processing scheme, the initial reinforcement learning model is updated to obtain a trained reinforcement learning model; Update the number of training times; Determine whether the updated training times are equal to the preset times; If the updated number of training times is less than the preset number of times, the next training data of the target training data in the training data sequence is used as the new target training data, and the trained reinforcement learning model is used as the new initial reinforcement learning model. The above steps are repeated until the updated number of training times is equal to the preset number of times, and the trained reinforcement learning model is used as the processing solution decision model.
7. The method according to claim 6, characterized in that If the training processing scheme is to maintain the status quo, the preset feedback function corresponding to the training processing scheme is: Wherein R1 represents the training feedback value, x represents the sum of the failure rate, CPU utilization, CPU load, memory usage, disk read rate, disk write rate and network throughput in the target training data, and b1, a, and m1 are preset values; If the training processing scheme is a backup, the preset feedback function corresponding to the training processing scheme is Wherein R2 represents the training feedback value, x represents the sum of the failure rate, CPU utilization, CPU load, memory usage, disk read rate, disk write rate and network throughput in the target training data, b2, a, m2 are preset values, and b2 is less than b1; If the training processing scheme is migration, the preset feedback function corresponding to the training processing scheme is Wherein R3 represents the training feedback value, x represents the sum of the failure rate, CPU utilization, CPU load, memory usage, disk read rate, disk write rate and network throughput in the target training data, b3, a, m3 are preset values, and b3 is less than b1.
8. A server processing solution determination device, characterized in that: include: A processing module, used to determine target performance parameters according to the acquired original performance parameters of the server; Prediction module for: Inputting the target performance parameter into a fault prediction model to obtain a predicted failure rate, wherein the fault prediction model is a pre-trained classification model that calculates the failure rate based on the performance parameter; Inputting the target performance parameter into a performance parameter prediction model to obtain a predicted performance parameter, wherein the performance parameter prediction model is a pre-trained time series model that calculates new performance parameters based on the input performance parameters; The processing module is also used to input the predicted failure rate and the predicted performance parameters into a processing solution decision model to obtain a target processing solution. The processing solution decision model is a pre-trained reinforcement learning model that determines the processing solution based on the failure rate and performance parameters. The target processing solution is one of maintaining the status quo, backing up, and migrating.
9. An electronic device, characterized in that: include: Processor, memory, communication interface; The memory is used to store executable instructions of the processor; The processor is configured to execute the server processing solution determination method according to any one of claims 1 to 7 by executing the executable instructions.
10. A readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the server processing solution determination method according to any one of claims 1 to 7 is implemented.