Cloud server operation and maintenance method and device based on large model, medium and equipment

By collecting, normalizing, and feature mapping cloud server performance data, and using large models to predict future anomalies, the problems of real-time monitoring and refined management in cloud server operation and maintenance are solved, improving prediction accuracy and operation and maintenance efficiency.

CN119048056APending Publication Date: 2024-11-29SHANDONG LANGCHAO YUNTOU INFORMATION TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411118507.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-15
Publication Date
2024-11-29

AI Technical Summary

Technical Problem

How to monitor the operational status of cloud servers in real time, achieve refined operation and maintenance management, and ensure that cloud servers operate stably and efficiently throughout their entire lifecycle has become a key issue that the industry urgently needs to address.

Method used

The system collects cloud server performance data at multiple time scales, performs normalization and feature mapping, integrates performance features from various time scales, inputs them into a pre-trained cloud server performance prediction model, and determines whether anomalies will occur within a preset future time period.

Benefits of technology

It improves the accuracy and efficiency of cloud server performance prediction, provides real-time and accurate status information for operations and maintenance personnel, optimizes operations and maintenance decisions, and improves overall operations and maintenance efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119048056A_ABST
    Figure CN119048056A_ABST
Patent Text Reader

Abstract

The invention provides a cloud server operation and maintenance method and device based on a large model, a medium and equipment. The method comprises the following steps: acquiring cloud server performance data under a plurality of time scales; performing normalization on the cloud server performance data under each time scale; performing feature mapping on the normalized cloud server performance data under each time scale to obtain cloud server performance features under the time scale; fusing the cloud server performance characteristics under each time scale to obtain cloud server performance fusion data; inputting the cloud server performance fusion data into a cloud server performance prediction large model to obtain cloud server performance prediction data in a future preset time period; and according to the cloud server performance prediction data, judging whether the cloud server is abnormal in a future preset time period. The method can provide real-time and accurate judgment results for operation and maintenance personnel, facilitates optimization of operation and maintenance decisions, and improves the overall operation and maintenance efficiency.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of operation and maintenance management technology, and in particular to a cloud server operation and maintenance method and device, medium, and equipment based on a large model. Background Art

[0002] With the rapid development of cloud computing technology, cloud server systems are becoming increasingly complex, their scale is constantly expanding, and their functions are becoming increasingly rich, which puts higher demands on operation and maintenance management. Facing this challenge, how to monitor the operating status of the server in real time, implement refined operation and maintenance management, and ensure that the cloud server runs stably and efficiently throughout its life cycle has become a key issue that the industry needs to solve urgently. Summary of the invention

[0003] In response to at least one of the above technical problems, an embodiment of the present invention provides a cloud server operation and maintenance method, device, medium, and equipment based on a large model.

[0004] According to the first aspect, the cloud server operation and maintenance method based on the large model provided by the embodiment of the present invention includes:

[0005] Collect cloud server performance data at multiple time scales;

[0006] Normalize the cloud server performance data at each time scale;

[0007] Perform feature mapping on the normalized cloud server performance data at each time scale to obtain the cloud server performance characteristics at that time scale; wherein the scale of the cloud server performance characteristics at each time scale is the same;

[0008] The cloud server performance characteristics at each time scale are integrated to obtain cloud server performance fusion data;

[0009] Inputting the cloud server performance fusion data into a pre-trained cloud server performance prediction model to obtain cloud server performance prediction data within a future preset time period;

[0010] According to the cloud server performance prediction data and the cloud server performance evaluation threshold, it is determined whether an abnormality occurs in the cloud server within the future preset time period.

[0011] In one embodiment, the cloud server performance data includes at least one of CPU usage, memory usage, and disk usage.

[0012] In one embodiment, the normalizing the cloud server performance data at each time scale includes: normalizing the cloud server performance data at each time scale using a first calculation formula, wherein the first calculation formula is:

[0013]

[0014] Among them, y is the cloud server performance data after normalization of x, x is the cloud server performance data at each time scale, E(x) represents the mean of the cloud server performance data at that time scale, VAR(x) represents the variance of the cloud server performance data at that time scale, ε is a very small number close to 0, which is the scaling parameter, and β is the translation parameter.

[0015] In one embodiment, the normalized cloud server performance data at each time scale is subjected to feature mapping to obtain the cloud server performance characteristics at the time scale, including:

[0016] The scale of the cloud server performance characteristics at each time scale is uniformly set to k×d;

[0017] The activation function of the hidden layer in the autoencoder is used to perform feature encoding on the normalized cloud server performance data at each time scale to obtain the cloud server performance features at that time scale; wherein the scale of the normalized cloud server performance data at each time scale is n×m;

[0018] Among them, the autoencoder includes an input layer, a hidden layer and an output layer, the input layer is used to receive the normalized cloud server performance data at each time scale, the hidden layer is used to encode the normalized cloud server performance data at each time scale into cloud server performance characteristics of a scale of k×d using an activation function; the output layer is used to decode the cloud server performance characteristics of a scale of k×d into data of a scale of n×m through an activation function; the mean square error between the input data of the input layer and the output data of the output layer is less than a preset value.

[0019] In one embodiment, judging whether an abnormality occurs in the cloud server within the future preset time period according to the cloud server performance prediction data and the cloud server performance evaluation threshold includes:

[0020] The cloud server performance prediction data is compared with the corresponding cloud server performance judgment threshold; if the cloud server performance prediction data is less than the corresponding cloud server performance judgment threshold, the performance of the cloud server is normal within the future preset time period; otherwise, the cloud server is abnormal within the future preset time period.

[0021] In one embodiment, the method further comprises:

[0022] If an abnormality occurs in the cloud server within the future preset time period, the potential factors causing the abnormality are analyzed based on the cloud server performance prediction data, and the analysis results of the potential factors are fed back to the operation and maintenance management personnel.

[0023] In one embodiment, the cloud server performance evaluation threshold includes at least one of a risk threshold of CPU utilization, a risk threshold of memory utilization, and a risk threshold of disk utilization; wherein the risk threshold of CPU utilization is within the range of (65%, 75%); wherein the risk threshold of memory utilization is within the range of (75%, 85%); wherein the risk threshold of disk utilization is within the range of (75%, 85%).

[0024] According to the second aspect, the cloud server operation and maintenance device based on the large model provided by the embodiment of the present invention includes:

[0025] A data collection module, used to collect cloud server performance data at multiple time scales;

[0026] A normalization processing module is used to normalize the cloud server performance data at each time scale;

[0027] A feature mapping module is used to perform feature mapping on the normalized cloud server performance data at each time scale to obtain the cloud server performance characteristics at the time scale; wherein the scale of the cloud server performance characteristics at each time scale is the same;

[0028] The feature fusion module is used to fuse the cloud server performance features at various time scales to obtain cloud server performance fusion data;

[0029] A performance prediction module, used to input the cloud server performance fusion data into a pre-trained cloud server performance prediction model to obtain cloud server performance prediction data within a future preset time period;

[0030] The abnormality judgment module is used to judge whether the cloud server has an abnormality within the future preset time period based on the cloud server performance prediction data and the cloud server performance judgment threshold.

[0031] According to a third aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed in a computer, enables the computer to execute the method provided in the first aspect.

[0032] According to a fourth aspect, an embodiment of the present invention provides a computing device, including a memory and a processor, wherein the memory stores executable code, and when the processor executes the executable code, the method provided in the first aspect is implemented.

[0033] The cloud server operation and maintenance method, device, medium, and equipment based on the large model provided in the embodiment of the present invention collects cloud server performance data at multiple time scales, normalizes the cloud server performance data at each time scale, and then performs feature mapping on the normalized cloud server performance data at each time scale to obtain the cloud server performance characteristics at the time scale, and the scale of the cloud server performance characteristics at each time scale is the same. Then, the cloud server performance characteristics at each time scale are fused to obtain cloud server performance fusion data, and the cloud server performance fusion data is input into the pre-trained cloud server performance prediction large model to obtain cloud server performance prediction data in the future preset time period, and then the cloud server performance prediction data is used to determine whether the cloud server is abnormal in the future preset time period. Since the present invention performs feature mapping on the server performance data at each time scale and fuses the server performance characteristics at each time scale, the fused multi-scale data can provide more useful information to the cloud server performance prediction large model, thereby improving the prediction accuracy and solving the problem of poor prediction model performance caused by multi-scale data in the prior art. Moreover, due to the introduction of large model technology, using its powerful data analysis and prediction capabilities, the accuracy and efficiency of cloud server performance prediction can be effectively improved, providing real-time and accurate status information for operation and maintenance personnel, thereby helping to optimize operation and maintenance decisions and improve overall operation and maintenance efficiency. In addition, since the embodiments of the present invention fully consider the cloud server performance data at multiple time scales, the constructed cloud server performance prediction large model has strong universality and good stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a flowchart of a cloud server operation and maintenance method based on a large model in one embodiment of the present invention;

[0035] Figure 2 The present invention is a block diagram of a cloud server operation and maintenance device based on a large model in one embodiment of the present invention. DETAILED DESCRIPTION

[0036] In a first aspect, an embodiment of the present invention provides a cloud server operation and maintenance method based on a large model, see Figure 1 The method comprises the following steps S110 to S160:

[0037] S110, collecting cloud server performance data at multiple time scales;

[0038] For example, one time scale is to collect cloud server performance data once every hour, one time scale is to collect cloud server performance data once every day, and one time scale is to collect cloud server performance data once every week.

[0039] The cloud server performance data may include at least one of CPU usage, memory usage and disk usage.

[0040] S120, normalizing the cloud server performance data at each time scale;

[0041] It is understandable that the cloud server performance data is normalized to facilitate the subsequent use of the data.

[0042] The normalizing of the cloud server performance data at each time scale in S120 may include: normalizing the cloud server performance data at each time scale using a first calculation formula, wherein the first calculation formula is:

[0043]

[0044] Among them, y is the cloud server performance data after normalization of x, x is the cloud server performance data at each time scale, E(x) represents the mean of the cloud server performance data at that time scale, VAR(x) represents the variance of the cloud server performance data at that time scale, ε is a very small number close to 0, which is the scaling parameter, and β is the translation parameter.

[0045] S130, performing feature mapping on the normalized cloud server performance data at each time scale to obtain the cloud server performance characteristics at the time scale; wherein the scales of the cloud server performance characteristics at each time scale are the same;

[0046] It can be seen that after feature mapping, the scale, that is, the size, of the cloud server performance characteristics at each time scale is consistent. For example, the cloud server performance characteristics at each time scale are all k×d feature matrices.

[0047] In one embodiment, the feature mapping of the normalized cloud server performance data at each time scale in S130 to obtain the cloud server performance feature at the time scale may include:

[0048] S131, uniformly setting the scale of cloud server performance characteristics at each time scale to k×d;

[0049] S132, using the activation function of the hidden layer in the autoencoder to feature encode the normalized cloud server performance data at each time scale to obtain the cloud server performance features at the time scale; wherein the scale of the normalized cloud server performance data at each time scale is n×m;

[0050] Among them, the autoencoder includes an input layer, a hidden layer and an output layer, the input layer is used to receive the normalized cloud server performance data at each time scale, the hidden layer is used to encode the normalized cloud server performance data at each time scale into cloud server performance characteristics of a scale of k×d using an activation function; the output layer is used to decode the cloud server performance characteristics of a scale of k×d into data of a scale of n×m through an activation function; the mean square error between the input data of the input layer and the output data of the output layer is less than a preset value.

[0051] The autoencoder includes an input layer, a hidden layer and an output layer. The function of the input layer is to receive the normalized cloud server performance data at each time scale. The function of the hidden layer is to encode the normalized cloud server performance data at each time scale into cloud server performance characteristics of a scale of k×d. In fact, the cloud server performance characteristics of a scale of k×d obtained by the hidden layer encoding are the cloud server performance characteristics we need. The purpose of setting the output layer is to reversely encode or decode the encoding of the above hidden layer to obtain data of a scale of n×m, and the scale of the data input to the input layer is also n×m. It can be seen that the scale of the input data of the input layer and the output data of the output layer is consistent. Then compare the input data of the input layer and the output data of the output layer. If the difference between the two is very small, it means that the encoding of the hidden layer is reasonable, otherwise the encoding process of the hidden layer needs to be adjusted.

[0052] For example, the hidden layer is encoded using the following function: h = f(w1 x + b1), and the output layer is decoded using the following function: w1 and w2 are weights, b1 and b2 are biases, the activation function f is the sigmoid function, and h is the encoding result of the hidden layer. is the decoding result of the output layer, and x is the input data of the input layer.

[0053] For another example, the loss function is used to characterize the gap between the input data of the input layer and the output data of the output layer. The loss function is:

[0054]

[0055] Among them, MSE is the loss value, m is the number of columns in the size n×m. The value of i is 1, 2, ..., m.

[0056] S140, integrating the cloud server performance characteristics at various time scales to obtain cloud server performance integration data;

[0057] Since the scale of cloud server performance characteristics at each time scale is consistent, it is very convenient to fuse the cloud server performance characteristics at each time scale. Specifically, the fusion of cloud server performance characteristics at each time scale can be achieved by matrix splicing, or by other methods, such as matrix addition.

[0058] S150, inputting the cloud server performance fusion data into a pre-trained cloud server performance prediction model to obtain cloud server performance prediction data within a future preset time period;

[0059] Among them, the cloud server performance prediction model is pre-trained using the performance fusion data of multiple cloud servers. The cloud server performance prediction model uses its powerful feature extraction and data analysis capabilities to learn and analyze the fused cloud server performance fusion data, fine-tune the transformer-based model, and obtain a cloud server performance prediction model that is more suitable for multiple scales. Using this cloud server performance prediction model for prediction, accurate server performance prediction results can be obtained.

[0060] Understandably, the emergence of big model technology provides a new perspective for people to process massive data. It simulates the cognitive process of the human brain by building a huge neural network, thereby realizing deep learning and processing of massive data. This neural network structure is extremely complex, but it is this complexity that enables it to adapt to various complex tasks. By continuously optimizing and adjusting the network structure, the big model can gradually approach the cognitive ability of humans, and even surpass humans in some aspects. With the support of big model technology, we can use large-scale data sets to train models. These data sets cover knowledge and information in various fields, providing rich learning resources for the model. Through continuous learning and optimization, the big model can gradually master various complex patterns and associations, thereby realizing accurate analysis and prediction of data. With the continuous increase of training data, the performance of the big model is gradually improved. It can more accurately identify, analyze and predict various phenomena, and provide more accurate and reliable intelligent services for humans. Whether in natural language processing, image processing or speech recognition, big models have shown great potential.

[0061] S160: Determine whether an abnormality occurs in the cloud server within the future preset time period based on the cloud server performance prediction data and the cloud server performance evaluation threshold.

[0062] It can be seen that after obtaining the cloud server performance prediction data, the cloud server performance prediction data can be compared with the cloud server performance judgment threshold to obtain a judgment result of whether the cloud server has an abnormality within the future preset time period.

[0063] In one embodiment, judging whether an abnormality occurs in the cloud server within the future preset time period according to the cloud server performance prediction data and the cloud server performance judgment threshold in S160 may include: comparing the cloud server performance prediction data with the corresponding cloud server performance judgment threshold; if the cloud server performance prediction data is less than the corresponding cloud server performance judgment threshold, the performance of the cloud server is normal within the future preset time period; otherwise, an abnormality occurs in the cloud server within the future preset time period.

[0064] That is, if the cloud server performance prediction data is less than the corresponding cloud server performance evaluation threshold, it is considered that the performance of the cloud server will most likely work normally in the future preset time period. If the cloud server performance prediction data is greater than or equal to the corresponding cloud server performance evaluation threshold, it is considered that the performance of the cloud server will most likely be abnormal in the future preset time period.

[0065] Furthermore, the method provided by the embodiment of the present invention may also include: if an abnormality occurs in the cloud server within the future preset time period, analyzing the potential factors causing the abnormality based on the cloud server performance prediction data, and feeding back the analysis results of the potential factors to the operation and maintenance management personnel.

[0066] That is to say, if it is judged that the performance of the cloud server is likely to be abnormal within a preset time period in the future, the cloud server performance prediction data is analyzed to find out the potential factors causing the abnormality, and the potential factors causing the abnormality are fed back to the operation and maintenance management personnel, so that the operation and maintenance management personnel can make auxiliary decisions in advance based on the feedback information, or conduct early troubleshooting.

[0067] The cloud server performance evaluation threshold includes at least one of a risk threshold of CPU utilization, a risk threshold of memory utilization, and a risk threshold of disk utilization; the risk threshold of CPU utilization is within the range of (65%, 75%); the risk threshold of memory utilization is within the range of (75%, 85%); the risk threshold of disk utilization is within the range of (75%, 85%).

[0068] For example, the risk threshold for CPU utilization is 70%, the risk threshold for memory utilization is 80%, and the risk threshold for disk utilization is 80%.

[0069] It is understandable that the embodiments of the present invention realize accurate prediction of server performance data at multiple time scales, and make judgments on whether an abnormality will occur based on the prediction results, thereby providing warning information with guiding suggestions. Since the present invention performs feature mapping on the server performance data at each time scale and fuses the server performance features at each time scale, the fused multi-scale data can provide more useful information to the cloud server performance prediction model, thereby improving the prediction accuracy and solving the problem of poor prediction model performance caused by multi-scale data in the prior art.

[0070] The embodiment of the present invention realizes the analysis of cloud server performance data under multiple time scales, improves the accuracy, stability and scope of application of the cloud server performance prediction model, optimizes operation and maintenance decisions, and improves the overall operation and maintenance efficiency. Moreover, due to the introduction of large model technology and its powerful data analysis and prediction capabilities, the accuracy and efficiency of cloud server performance prediction can be effectively improved, providing real-time and accurate status information for operation and maintenance personnel, thereby helping to optimize operation and maintenance decisions and improve overall operation and maintenance efficiency.

[0071] The embodiment of the present invention can integrate cloud server performance data at multiple time scales, improve the prediction accuracy and learning ability of the cloud server performance prediction model, and be used to implement the operation and maintenance management of the cloud server, ensure the safe and stable operation of the cloud server, and improve the operation and maintenance management efficiency of the cloud server. Since the embodiment of the present invention fully considers the cloud server performance data at multiple time scales, the constructed cloud server performance prediction model has strong universality and good stability.

[0072] In a second aspect, an embodiment of the present invention provides a cloud server operation and maintenance device based on a large model, see Figure 2 , the device 100 comprises:

[0073] A data collection module 110 is used to collect cloud server performance data at multiple time scales;

[0074] A normalization processing module 120 is used to perform normalization processing on the cloud server performance data at each time scale;

[0075] The feature mapping module 130 is used to perform feature mapping on the normalized cloud server performance data at each time scale to obtain the cloud server performance characteristics at the time scale; wherein the scale of the cloud server performance characteristics at each time scale is the same;

[0076] The feature fusion module 140 is used to fuse the cloud server performance features at various time scales to obtain cloud server performance fusion data;

[0077] The performance prediction module 150 is used to input the cloud server performance fusion data into a pre-trained cloud server performance prediction model to obtain cloud server performance prediction data within a future preset time period;

[0078] The abnormality judgment module 160 is used to judge whether the cloud server has an abnormality within the future preset time period according to the cloud server performance prediction data and the cloud server performance judgment threshold.

[0079] In one embodiment, the cloud server performance data includes at least one of CPU usage, memory usage, and disk usage.

[0080] In one embodiment, the normalization processing module is specifically used to: use a first calculation formula to normalize the cloud server performance data at each time scale, and the first calculation formula is:

[0081]

[0082] Among them, y is the cloud server performance data after normalization of x, x is the cloud server performance data at each time scale, E(x) represents the mean of the cloud server performance data at that time scale, VAR(x) represents the variance of the cloud server performance data at that time scale, ε is a very small number close to 0, which is the scaling parameter, and β is the translation parameter.

[0083] In one embodiment, the feature mapping module is specifically used to: uniformly set the scale of cloud server performance characteristics at each time scale to k×d; use the activation function of the hidden layer in the autoencoder to feature encode the normalized cloud server performance data at each time scale to obtain the cloud server performance characteristics at the time scale; wherein the scale of the normalized cloud server performance data at each time scale is n×m; wherein the autoencoder includes an input layer, a hidden layer and an output layer, the input layer is used to receive the normalized cloud server performance data at each time scale, the hidden layer is used to encode the normalized cloud server performance data at each time scale into a cloud server performance characteristic of a scale of k×d using an activation function; the output layer is used to decode the cloud server performance characteristic of a scale of k×d into data of a scale of n×m through an activation function; the mean square error between the input data of the input layer and the output data of the output layer is less than a preset value.

[0084] In one embodiment, the abnormality judgment module is specifically used to: compare the cloud server performance prediction data with the corresponding cloud server performance judgment threshold; if the cloud server performance prediction data is less than the corresponding cloud server performance judgment threshold, the performance of the cloud server is normal within the future preset time period; otherwise, an abnormality occurs in the cloud server within the future preset time period.

[0085] In one embodiment, the device may further include:

[0086] The exception handling module is used to analyze the potential factors causing the abnormality based on the cloud server performance prediction data if an abnormality occurs in the cloud server within the preset time period in the future, and feed back the analysis results of the potential factors to the operation and maintenance management personnel.

[0087] In one embodiment, the cloud server performance evaluation threshold includes at least one of a risk threshold of CPU utilization, a risk threshold of memory utilization, and a risk threshold of disk utilization; wherein the risk threshold of CPU utilization is within the range of (65%, 75%); wherein the risk threshold of memory utilization is within the range of (75%, 85%); wherein the risk threshold of disk utilization is within the range of (75%, 85%).

[0088] It is understandable that the explanation, specific implementation, beneficial effects, examples, etc. of the relevant contents in the device provided in the embodiment of the present invention can be found in the corresponding parts of the method provided in the first aspect, and will not be repeated here.

[0089] In a third aspect, an embodiment of the present invention provides a computer-readable medium, on which computer instructions are stored. When the computer instructions are executed by a processor, the processor executes the method provided in the first aspect.

[0090] Specifically, a system or device equipped with a storage medium can be provided, on which software program code that implements the functions of any of the above-mentioned embodiments is stored, and a computer (or CPU or MPU) of the system or device can be enabled to read and execute the program code stored in the storage medium.

[0091] In this case, the program code itself read from the storage medium can realize the function of any one of the above-mentioned embodiments, and thus the program code and the storage medium storing the program code constitute a part of the present invention.

[0092] The storage medium embodiments for providing the program code include a floppy disk, a hard disk, a magneto-optical disk, an optical disk (such as CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RAM, DVD-RW, DVD+RW), a magnetic tape, a non-volatile memory card, and a ROM. Alternatively, the program code can be downloaded from a server computer by a communication network.

[0093] In addition, it should be clear that the functions of any of the above embodiments can be implemented not only by executing the program code read by the computer, but also by enabling an operating system operating on the computer to complete part or all of the actual operations based on instructions from the program code.

[0094] In addition, it can be understood that the program code read from the storage medium is written to a memory provided in an expansion board inserted into the computer or to a memory provided in an expansion module connected to the computer, and then based on the instructions of the program code, a CPU installed on the expansion board or expansion module is enabled to perform part or all of the actual operations, thereby realizing the functions of any of the above-mentioned embodiments.

[0095] It is understandable that the explanation, specific implementation methods, beneficial effects, examples, etc. of the relevant contents in the computer-readable medium provided in the embodiment of the present invention can be found in the corresponding parts of the method provided in the first aspect, and will not be repeated here.

[0096] In a fourth aspect, an embodiment of the present specification provides a computing device, including a memory and a processor, wherein the memory stores executable code, and when the processor executes the executable code, it implements the method in any embodiment of the specification.

[0097] It is understandable that the explanation, specific implementation, beneficial effects, examples, etc. of the relevant contents in the computing device provided in the embodiment of the present invention can be found in the corresponding parts of the method provided in the first aspect, and will not be repeated here.

[0098] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0099] Those skilled in the art should be aware that in one or more of the above examples, the functions described in the present invention can be implemented by hardware, software, widgets, or any combination thereof. When implemented by software, these functions can be stored in a computer-readable medium or transmitted as one or more instructions or codes on a computer-readable medium.

[0100] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made on the basis of the technical solution of the present invention should be included in the scope of protection of the present invention.

Claims

1. A cloud server operation and maintenance method based on a large model, characterized in that: include: Collect cloud server performance data at multiple time scales; Normalize the cloud server performance data at each time scale; Perform feature mapping on the normalized cloud server performance data at each time scale to obtain the cloud server performance characteristics at that time scale; wherein the scale of the cloud server performance characteristics at each time scale is the same; The cloud server performance characteristics at each time scale are integrated to obtain cloud server performance fusion data; Inputting the cloud server performance fusion data into a pre-trained cloud server performance prediction model to obtain cloud server performance prediction data within a future preset time period; According to the cloud server performance prediction data and the cloud server performance evaluation threshold, it is determined whether an abnormality occurs in the cloud server within the future preset time period.

2. The method according to claim 1, characterized in that: The cloud server performance data includes at least one of CPU usage, memory usage, and disk usage.

3. The method according to claim 1, characterized in that The normalizing the cloud server performance data at each time scale includes: using a first calculation formula to normalize the cloud server performance data at each time scale, wherein the first calculation formula is: Among them, y is the cloud server performance data after normalization of x, x is the cloud server performance data at each time scale, E(x) represents the mean of the cloud server performance data at that time scale, VAR(x) represents the variance of the cloud server performance data at that time scale, ε is a very small number close to 0, which is the scaling parameter, and β is the translation parameter.

4. The method according to claim 1, characterized in that: The normalized cloud server performance data at each time scale is subjected to feature mapping to obtain the cloud server performance characteristics at the time scale, including: The scale of the cloud server performance characteristics at each time scale is uniformly set to k×d; The activation function of the hidden layer in the autoencoder is used to perform feature encoding on the normalized cloud server performance data at each time scale to obtain the cloud server performance features at that time scale; wherein the scale of the normalized cloud server performance data at each time scale is n×m; Among them, the autoencoder includes an input layer, a hidden layer and an output layer, the input layer is used to receive the normalized cloud server performance data at each time scale, the hidden layer is used to encode the normalized cloud server performance data at each time scale into cloud server performance characteristics of a scale of k×d using an activation function; the output layer is used to decode the cloud server performance characteristics of a scale of k×d into data of a scale of n×m through an activation function; the mean square error between the input data of the input layer and the output data of the output layer is less than a preset value.

5. The method according to claim 1, characterized in that The determining, based on the cloud server performance prediction data and the cloud server performance evaluation threshold, whether the cloud server is abnormal within the future preset time period includes: The cloud server performance prediction data is compared with the corresponding cloud server performance judgment threshold; if the cloud server performance prediction data is less than the corresponding cloud server performance judgment threshold, the performance of the cloud server is normal within the future preset time period; otherwise, the cloud server is abnormal within the future preset time period.

6. The method according to claim 1, characterized in that Also includes: If an abnormality occurs in the cloud server within the future preset time period, the potential factors causing the abnormality are analyzed based on the cloud server performance prediction data, and the analysis results of the potential factors are fed back to the operation and maintenance management personnel.

7. The method according to claim 1, characterized in that The cloud server performance evaluation threshold includes at least one of a risk threshold of CPU utilization, a risk threshold of memory utilization, and a risk threshold of disk utilization; wherein the risk threshold of CPU utilization is within the range of (65%, 75%); wherein the risk threshold of memory utilization is within the range of (75%, 85%); wherein the risk threshold of disk utilization is within the range of (75%, 85%).

8. A cloud server operation and maintenance device based on a large model, characterized in that: include: A data collection module, used to collect cloud server performance data at multiple time scales; A normalization processing module is used to normalize the cloud server performance data at each time scale; A feature mapping module is used to perform feature mapping on the normalized cloud server performance data at each time scale to obtain the cloud server performance characteristics at the time scale; wherein the scale of the cloud server performance characteristics at each time scale is the same; The feature fusion module is used to fuse the cloud server performance features at various time scales to obtain cloud server performance fusion data; A performance prediction module, used to input the cloud server performance fusion data into a pre-trained cloud server performance prediction model to obtain cloud server performance prediction data within a future preset time period; The abnormality judgment module is used to judge whether the cloud server has an abnormality within the future preset time period based on the cloud server performance prediction data and the cloud server performance judgment threshold.

9. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed in a computer, the computer is caused to execute the method according to any one of claims 1 to 7.

10. A computing device, characterized in that The method comprises a memory and a processor, wherein the memory stores executable codes, and when the processor executes the executable codes, the method according to any one of claims 1 to 7 is implemented.