Server performance detection method based on deep learning
By using a deep learning-based server performance testing method, which leverages LSTM networks and data encryption technology, the problems of long testing cycles and poor accuracy in traditional methods are solved, enabling efficient and intelligent management of server performance and fault prediction.
Patent Information
- Application Number
- CN202510988720.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-17
- Publication Date
- 2025-11-04
AI Technical Summary
Traditional server performance testing methods struggle to handle multi-dimensional and multi-modal data and lack intelligent management, resulting in long testing cycles, poor accuracy, and an inability to detect potential faults in a timely manner.
By employing a deep learning-based approach, a deep learning model is trained by collecting multi-dimensional performance data. This model is then combined with an LSTM network to establish a performance degradation prediction model. The system dynamically collects GPU latency, power consumption, and memory usage, identifies abnormal behavior, and performs data encryption and anonymization. This results in the construction of a performance prediction and anomaly detection module, enabling adaptive load control and automated early warning.
It enables efficient and accurate detection of server performance, provides early warning of hardware aging, dynamically adjusts model complexity, improves intelligent management, and reduces detection cycles and the time required to discover potential faults.
Smart Images
Figure CN120892302A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of server performance testing technology, specifically a server performance testing method based on deep learning. Background Technology
[0002] With the rapid development of information technology, enterprises are increasingly reliant on servers. As crucial hardware supporting daily business operations and data processing, the stability and reliability of server performance directly impact operational efficiency and data security. However, during prolonged high-load operation, servers frequently encounter various performance issues, such as CPU overload, memory leaks, and disk I / O bottlenecks. These performance problems can lead to serious consequences such as server crashes, slow response times, and data loss, causing significant losses to enterprises. As server hardware performance continues to improve, how to efficiently and accurately monitor server performance has become a critical issue for enterprise data center operation and maintenance management. Traditional performance testing methods are mostly based on manually setting hardware performance indicators and rule analysis. However, in practical applications, these methods struggle to cope with the multi-dimensional complexity and dynamic changes in server performance.
[0003] In recent years, with the development of big data and machine learning technologies, performance monitoring methods based on statistical analysis and machine learning have been widely used. However, these methods still have some limitations: Traditional machine learning methods typically require extensive manual feature engineering and preprocessing steps, and still struggle to achieve efficient and accurate performance prediction when faced with multi-dimensional and multi-modal data. Staff are unable to obtain real-time operating data for every step and every piece of electrical equipment, resulting in a low overall level of intelligent management. The performance evaluation and prediction capabilities of automation are insufficient to meet the needs of normal testing, which can easily lead to a longer overall testing cycle.
[0004] Therefore, this invention requires the design of a server performance testing method based on deep learning to solve the aforementioned problems. Summary of the Invention The purpose of this invention is to provide a server performance testing method based on deep learning to solve the above-mentioned problems, thereby addressing the issues mentioned in the background section.
[0005] To address the above problems, the present invention provides a technical solution: A deep learning-based server performance testing method includes the following specific steps: S1. Collect multi-dimensional performance data from the server to ensure that a compatible GPU driver can be deployed in a bare-metal Linux environment; S2. Train a deep learning model using historical performance data, and then load a pre-built computing power test model in conjunction with the deep learning framework; S3. Make predictions based on historical performance data and deep learning models, and dynamically collect the latency, power consumption, memory usage and error rate of the GPU during the model inference process; S4. Establish a performance degradation prediction model based on LSTM network, output a computing power stability report, convert server performance data into a format suitable for deep learning model analysis, and convert the original server performance data into a visualization image or feature map.
[0006] In a preferred embodiment of the present invention, the performance data in step S1 includes CPU load, memory usage, disk I / O, and network latency.
[0007] In a preferred embodiment of the present invention, after dynamic data collection is completed in step S3, it is also necessary to identify abnormal behaviors in the server performance data, promptly detect potential faults or abnormal resource consumption, and once an anomaly is detected, the anomaly detection unit can further analyze possible causes and promptly conduct on-site early warning processing when it cannot resolve the issue on its own.
[0008] In a preferred embodiment of the present invention, when establishing the performance degradation prediction model in step S4, data encryption and anonymization are also required to avoid leaking sensitive information. At the same time, privacy keywords and warning keywords can be preset to respond accordingly when relevant keywords appear.
[0009] In a preferred embodiment of the present invention, the deep learning model in step S2 includes a convolutional neural network (CNN) and a long short-term memory network (LSTM).
[0010] In a preferred embodiment of the present invention, a server performance testing system needs to be constructed before performing step S1. The server performance testing system includes a performance prediction and anomaly detection module, a deep learning model, a server privacy and security module, and a data acquisition and preprocessing module. The output of the data acquisition and preprocessing module is communicatively connected to the input of the deep learning model, and the output of the deep learning model is communicatively connected to the input of the performance prediction and anomaly detection module. The server privacy and security module is integrated into the performance prediction and anomaly detection module, the deep learning model, and the data acquisition and preprocessing module.
[0011] In a preferred embodiment of the present invention, the performance prediction and anomaly detection module includes a performance prediction unit, an anomaly detection module unit, and an on-site early warning unit. The output terminal of the performance prediction unit is communicatively connected to the input terminal of the anomaly detection module unit, and the on-site early warning unit is integrated inside the anomaly detection module unit. The performance prediction unit is used to predict and assess server performance changes in advance based on historical performance data and deep learning models. The anomaly detection unit is used to dynamically collect the latency, power consumption, memory usage and error rate of the GPU during the model inference process, establish a performance degradation prediction model based on the LSTM network, output a computing power stability report, identify abnormal behavior in server performance data, and promptly detect potential faults or abnormal resource consumption. Once an anomaly is detected, the anomaly detection unit can further analyze possible causes and promptly provide on-site early warning when it cannot resolve the issue on its own. The on-site early warning unit is used to issue timely on-site alarms or take automated emergency measures based on the results of performance prediction and anomaly detection.
[0012] In a preferred embodiment of the present invention, the deep learning model includes a learning model training and optimization unit, an algorithm optimization database, and an algorithm access unit. The output end of the algorithm access unit is communicatively connected to the input end of the algorithm optimization database, and the learning model training and optimization unit is bidirectionally communicatively connected to the algorithm optimization database. The learning model training and optimization unit is used to train a deep learning model using historical performance data, optimize the model's accuracy and robustness, and load a pre-set computing power test model through a deep learning framework. It is also used to fine-tune models for different server environments using methods such as transfer learning; The algorithm-optimized database is used to analyze and predict server performance through deep learning models, identify patterns in time series data based on deep learning models, and help select the features that best characterize performance changes. It is also used to accelerate the training and inference process of models using GPUs or TPUs; The algorithm access unit is used to continuously add real-time updated algorithms to the algorithm database, thereby improving the comprehensiveness of subsequent calculations.
[0013] In a preferred embodiment of the present invention, the server privacy and security module includes a local server control unit, a privacy and security monitoring unit, and a data imaging unit. The output end of the local server control unit is communicatively connected to the input end of the data imaging unit, and the privacy and security monitoring unit is integrated inside the local server control unit. The local server control unit is used to manage and control the server's performance testing process, ensure accurate data collection and processing, interact with the deep learning model, periodically upload or transmit the collected performance data to the server for analysis, and also make corresponding controls and adjustments based on the feedback from the deep learning model. The privacy and security monitoring unit is used to encrypt and anonymize data to prevent the leakage of sensitive information. It can also pre-set privacy keywords and warning keywords and react accordingly when relevant keywords appear. The data imaging unit is used to convert server performance data into a format suitable for deep learning model analysis and to provide data support for further model training and prediction, transforming the original server performance data into a visual image or feature map.
[0014] In a preferred embodiment of the present invention, the data acquisition and preprocessing module includes a data acquisition unit, a data preprocessing unit, and a feature extraction and selection unit. The output end of the data acquisition unit is communicatively connected to the input end of the data preprocessing unit, and the output end of the data preprocessing unit is communicatively connected to the input end of the feature extraction and selection unit. The acquisition unit is used to collect multi-dimensional performance data from the server, including CPU load, memory usage, disk I / O, and network latency; The data preprocessing unit is used to perform preprocessing on the data, such as standardization, missing value imputation, and outlier detection, to ensure data quality. The feature extraction and selection unit is used to automatically extract key features from raw data using autoencoder methods in deep learning, and to deploy an adapted GPU driver in a bare-metal Linux environment.
[0015] The beneficial effects of this invention are as follows: By setting up a performance prediction and anomaly detection module, a deep learning model, a server privacy and security module, and a data acquisition and preprocessing module, this invention constructs a comprehensive server performance testing system. In actual operation, it collects multi-dimensional performance data from the server, ensuring that a compatible GPU driver can be deployed in a bare-metal Linux environment. A deep learning model is trained using historical performance data, thereby cooperating with a pre-built computing power test model loaded within a deep learning framework. Predictions are made based on historical performance data and the deep learning model, dynamically collecting latency, power consumption, memory usage, and error rate of the GPU during model inference. A performance degradation prediction model is established based on an LSTM network, outputting a computing power stability report, and converting server performance data into suitable... The system adopts a format compatible with deep learning model analysis, transforming raw server performance data into visual images or feature maps. This facilitates the management and processing of each module, unit, and power device. By using an LSTM model to predict the upward trend of memory error rate, it provides early warning of hardware aging, enabling fault prediction and handling. The system dynamically adjusts model complexity to adapt to the extreme testing requirements of different GPU specifications, achieving adaptive load. The introduction of deep learning can effectively analyze massive amounts of performance data, extract valuable features, and perform automated performance evaluation and prediction. It manages, visualizes, and stores server performance testing data and corresponding analysis results, which helps to realize server performance testing management through IoT cloud control and improve the level of intelligence in server performance testing management. Attached Figure Description For ease of explanation, the present invention will be described in detail below with reference to specific embodiments and accompanying drawings.
[0016] Figure 1 This is an overall flowchart of a deep learning-based server performance testing method according to the present invention. Detailed Implementation like Figure 1 As shown, the specific implementation adopts the following technical solution: A deep learning-based server performance testing method includes the following specific steps: S1. Collect multi-dimensional performance data from the server to ensure that a compatible GPU driver can be deployed in a bare-metal Linux environment; Performance data includes CPU load, memory usage, disk I / O, and network latency; S2. Train a deep learning model using historical performance data, and then load a pre-built computing power test model in conjunction with the deep learning framework; Deep learning models include Convolutional Neural Networks (CNNs) and Long Short-Term Memory Networks (LSTMs). S3. Make predictions based on historical performance data and deep learning models, and dynamically collect the latency, power consumption, memory usage and error rate of the GPU during the model inference process; After dynamic data collection is completed, it is also necessary to identify abnormal behaviors in server performance data, promptly detect potential faults or abnormal resource consumption, and once an anomaly is detected, the anomaly detection unit can further analyze possible causes and promptly issue on-site warnings when it cannot resolve the issue on its own. S4. Establish a performance degradation prediction model based on LSTM network, output computing power stability report, convert server performance data into a format suitable for deep learning model analysis, and convert the original server performance data into a visualization image or feature map. When establishing a performance degradation prediction model, data encryption and anonymization are also required to avoid leaking sensitive information. At the same time, privacy keywords and warning keywords can be pre-set, and corresponding reactions can be made when relevant keywords appear.
[0017] Furthermore, before proceeding to step S1, a server performance testing system needs to be constructed. The server performance testing system includes a performance prediction and anomaly detection module, a deep learning model, a server privacy and security module, and a data acquisition and preprocessing module. The output of the data acquisition and preprocessing module is communicatively connected to the input of the deep learning model, and the output of the deep learning model is communicatively connected to the input of the performance prediction and anomaly detection module. The server privacy and security module is integrated into the performance prediction and anomaly detection module, the deep learning model, and the data acquisition and preprocessing module, respectively.
[0018] Furthermore, the performance prediction and anomaly detection module includes a performance prediction unit, an anomaly detection unit, and an on-site early warning unit. The output of the performance prediction unit is communicatively connected to the input of the anomaly detection unit, and the on-site early warning unit is integrated within the anomaly detection unit. The performance prediction unit is used to make predictions based on historical performance data and deep learning models to assess server performance changes in advance over a future period. The anomaly detection unit is used to dynamically collect GPU latency, power consumption, memory usage, and error rate during model inference, establish a performance degradation prediction model based on an LSTM network, output a computing power stability report, identify abnormal behaviors in server performance data, and promptly detect potential faults or abnormal resource consumption. Once an anomaly is detected, the anomaly detection unit can further analyze possible causes and promptly issue on-site early warnings when it cannot resolve the issue itself. The on-site early warning unit is used to issue on-site alarms or take automated emergency measures in a timely manner based on the results of performance prediction and anomaly detection.
[0019] Furthermore, the deep learning model includes a learning model training and optimization unit, an algorithm optimization database, and an algorithm access unit. The output of the algorithm access unit is communicatively connected to the input of the algorithm optimization database, and the learning model training and optimization unit is bidirectionally communicatively connected to the algorithm optimization database. The learning model training and optimization unit is used to train the deep learning model using historical performance data, optimize the model's accuracy and robustness, and load a pre-set computing power test model through the deep learning framework. It is also used to fine-tune the model for different server environments using methods such as transfer learning. The algorithm optimization database is used to analyze and predict server performance through the deep learning model, identify patterns in time series data based on the deep learning model, and help select the features that best represent performance changes. It is also used to accelerate the model's training and inference process using GPUs or TPUs. The algorithm access unit is used to continuously add real-time updated algorithms to the algorithm database, thereby improving the comprehensiveness of subsequent calculations.
[0020] Furthermore, the server privacy and security module includes a local server control unit, a privacy and security monitoring unit, and a data imaging unit. The output of the local server control unit is communicatively connected to the input of the data imaging unit, and the privacy and security monitoring unit is integrated within the local server control unit. The local server control unit manages and controls the server's performance testing process, ensuring accurate data collection and processing, and interacts with the deep learning model to periodically upload or transmit the collected performance data to the server for analysis. It can also perform corresponding controls and adjustments based on feedback from the deep learning model. The privacy and security monitoring unit performs data encryption and anonymization to prevent the leakage of sensitive information. It can also pre-set privacy keywords and warning keywords to react accordingly when relevant keywords appear. The data imaging unit converts server performance data into a format suitable for deep learning model analysis and provides data support for further model training and prediction, transforming the raw server performance data into visual images or feature maps.
[0021] Furthermore, the data acquisition and preprocessing module includes a data acquisition unit, a data preprocessing unit, and a feature extraction and selection unit. The output of the data acquisition unit is communicatively connected to the input of the data preprocessing unit, and the output of the data preprocessing unit is communicatively connected to the input of the feature extraction and selection unit. The acquisition unit is used to collect multi-dimensional performance data from the server, including CPU load, memory usage, disk I / O, and network latency. The data preprocessing unit is used to perform preprocessing on the data, such as standardization, missing value imputation, and outlier detection, to ensure data quality. The feature extraction and selection unit is used to automatically extract key features from the raw data using autoencoder methods in deep learning and deploy an adapted GPU driver in a bare-metal Linux environment.
[0022] Example When deep learning methods are connected to server performance monitoring and management: S1. After staff check the on-site and remote power equipment one by one and confirm that they are all in normal operation, the server performance testing system is started. The server performance testing system controls the acquisition unit to collect multi-dimensional performance data from the server, including CPU load, memory usage, disk I / O, and network latency. The server performance testing system controls the data preprocessing unit to perform preprocessing such as standardization, missing value imputation, and outlier detection to ensure data quality. The server performance testing system controls the feature extraction and selection unit to automatically extract key features from the raw data using the autoencoder method in deep learning and deploy the adapted GPU driver in the bare-metal Linux environment. S2. The server performance detection system controls the performance prediction unit to make predictions based on historical performance data and deep learning models, and to assess the server's performance changes in the future. The server performance detection system controls the anomaly detection unit to dynamically collect the GPU's latency, power consumption, memory usage and error rate during the model inference process, establish a performance degradation prediction model based on the LSTM network, output a computing power stability report, identify abnormal behavior in server performance data, and promptly detect potential faults or abnormal resource consumption. Once an anomaly is detected, the anomaly detection unit can further analyze possible causes and promptly issue on-site warnings when it cannot resolve the issue on its own. The server performance detection system controls the on-site warning unit to issue on-site alarms or take automated emergency measures in a timely manner based on the performance prediction and anomaly detection results. S3, the server performance detection system control learning model training and optimization unit trains deep learning models using historical performance data, optimizes the accuracy and robustness of the models, loads pre-built computing power test models through deep learning frameworks, and also uses methods such as transfer learning to fine-tune the models for different server environments. The server performance detection system control algorithm optimization database analyzes and predicts server performance through deep learning models, identifies patterns in time series data based on deep learning models, and helps select the features that best represent performance changes. The server performance detection system control utilizes GPUs or TPUs to accelerate the model training and inference process, and the server performance detection system control algorithm access unit continuously adds real-time updated algorithms to the algorithm database to improve the comprehensiveness of subsequent calculations. S4, the server performance monitoring system controls the local server control unit to manage and control the server's performance monitoring process, ensuring accurate data collection and processing. It interacts with the deep learning model, periodically uploading or transmitting the collected performance data to the server for analysis. Based on feedback from the deep learning model, it can also make corresponding controls and adjustments. The server performance monitoring system's privacy and security monitoring unit encrypts and anonymizes data to prevent the leakage of sensitive information. It can also pre-set privacy keywords and warning keywords, reacting accordingly when relevant keywords appear. The server performance monitoring system's data imaging unit converts server performance data into a format suitable for deep learning model analysis, providing data support for further model training and prediction, and transforming the raw server performance data into visual images or feature maps.
[0023] Specifically, in practical applications, multiple data acquisition and preprocessing modules are used in conjunction with performance prediction and anomaly detection modules, deep learning models, and server privacy and security modules. These modules are located in different geographical locations. This invention constructs a comprehensive server performance monitoring system by setting up performance prediction and anomaly detection modules, deep learning models, server privacy and security modules, and data acquisition and preprocessing modules. In actual operation, multi-dimensional performance data is collected from the server, ensuring that a compatible GPU driver can be deployed in a bare-metal Linux environment. A deep learning model is trained using historical performance data, thereby loading a pre-built computing power testing model in conjunction with the deep learning framework. Predictions are made based on historical performance data and the deep learning model, dynamically collecting GPU latency, power consumption, memory usage, and error rate during model inference. LSTM networks establish performance degradation prediction models, output computing power stability reports, convert server performance data into a format suitable for deep learning model analysis, and transform raw server performance data into visual images or feature maps, facilitating the management and processing of each module, unit, and power device. LSTM models predict the upward trend of memory error rates, providing early warnings of hardware aging and enabling fault prediction and handling. Dynamically adjusting model complexity to adapt to the extreme testing requirements of different GPU specifications achieves adaptive load. The introduction of deep learning can effectively analyze massive amounts of performance data, extract valuable features, and perform automated performance evaluation and prediction. Managing, visualizing, and storing server performance testing data and corresponding analysis results helps to achieve server performance testing management through IoT cloud control, improving the intelligence level of server performance testing management.
[0024] Those skilled in the art will recognize that the modules and method steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.
[0025] In the embodiments provided in this application, it should be understood that the disclosed devices, systems, and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or units may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or equipment, and may be electrical, mechanical, or other forms.
[0026] The modules for performance prediction and anomaly detection, deep learning models, server privacy and security, and data acquisition and preprocessing may or may not be physically separate. The components displayed as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.
[0027] In addition, the functional modules in the various embodiments of the present invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.
[0028] It should be noted that the above examples are merely specific embodiments of the present invention, and the present invention is obviously not limited to the above embodiments, with many similar variations. All modifications that can be directly derived or conceived by those skilled in the art from the content disclosed in this invention should fall within the protection scope of this invention.
[0029] The above are merely preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.
Claims
1. A server performance testing method based on deep learning, characterized in that, The specific steps include the following: S1. Collect multi-dimensional performance data from the server to ensure that a compatible GPU driver can be deployed in a bare-metal Linux environment; S2. Train a deep learning model using historical performance data, and then load a pre-built computing power test model in conjunction with the deep learning framework; S3. Make predictions based on historical performance data and deep learning models, and dynamically collect the latency, power consumption, memory usage and error rate of the GPU during the model inference process; S4. Establish a performance degradation prediction model based on LSTM network, output a computing power stability report, convert server performance data into a format suitable for deep learning model analysis, and convert the original server performance data into a visualization image or feature map.
2. The server performance testing method based on deep learning according to claim 1, characterized in that: The performance data in step S1 includes CPU load, memory usage, disk I / O, and network latency.
3. The server performance testing method based on deep learning according to claim 1, characterized in that: In step S3, after dynamic data collection is completed, it is also necessary to identify abnormal behaviors in the server performance data, promptly detect potential faults or abnormal resource consumption, and once an anomaly is detected, the anomaly detection unit can further analyze possible causes and promptly issue on-site warnings when it cannot resolve the issue on its own.
4. The server performance testing method based on deep learning according to claim 3, characterized in that: In step S4, when establishing the performance degradation prediction model, data encryption and anonymization are also required to avoid leaking sensitive information. At the same time, privacy keywords and warning keywords can be preset to respond accordingly when relevant keywords appear.
5. The server performance testing method based on deep learning according to claim 1, characterized in that: The deep learning models in step S2 include Convolutional Neural Network (CNN) and Long Short-Term Memory Network (LSTM).
6. The server performance testing method based on deep learning according to claim 5, characterized in that: Before proceeding to step S1, a server performance testing system needs to be constructed. The server performance testing system includes a performance prediction and anomaly detection module, a deep learning model, a server privacy and security module, and a data acquisition and preprocessing module. The output of the data acquisition and preprocessing module is communicatively connected to the input of the deep learning model, and the output of the deep learning model is communicatively connected to the input of the performance prediction and anomaly detection module. The server privacy and security module is integrated into the performance prediction and anomaly detection module, the deep learning model, and the data acquisition and preprocessing module.
7. The server performance testing method based on deep learning according to claim 6, characterized in that: The performance prediction and anomaly detection module includes a performance prediction unit, an anomaly detection module unit, and an on-site early warning unit. The output of the performance prediction unit is communicatively connected to the input of the anomaly detection module unit, and the on-site early warning unit is integrated inside the anomaly detection module unit. The performance prediction unit is used to make predictions based on historical performance data and deep learning models; The anomaly detection unit is used to dynamically collect the latency, power consumption, memory usage and error rate of the GPU during the model inference process, and to establish a performance degradation prediction model based on the LSTM network. The on-site early warning unit is used to predict performance and detect anomalies based on the results.
8. The server performance testing method based on deep learning according to claim 6, characterized in that: The deep learning model includes a learning model training and optimization unit, an algorithm optimization database, and an algorithm access unit. The output of the algorithm access unit is communicatively connected to the input of the algorithm optimization database, and the learning model training and optimization unit is bidirectionally communicatively connected to the algorithm optimization database. The learning model training and optimization unit is used to train a deep learning model using historical performance data. The algorithm-optimized database is used to analyze and predict server performance through deep learning models; The algorithm access unit is used to continuously add real-time updated algorithms to the algorithm database.
9. The server performance testing method based on deep learning according to claim 6, characterized in that: The server privacy and security module includes a local server control unit, a privacy and security monitoring unit, and a data imaging unit. The output of the local server control unit is communicatively connected to the input of the data imaging unit, and the privacy and security monitoring unit is integrated inside the local server control unit. The local server control unit is used to manage and control the server's performance detection process, and interacts with the deep learning model to periodically upload or transmit the collected performance data to the server for analysis. The privacy and security monitoring unit is used to perform data encryption and anonymization to prevent the leakage of sensitive information; The data imaging unit is used to convert server performance data into a format suitable for deep learning model analysis.
10. The server performance testing method based on deep learning according to claim 6, characterized in that: The data acquisition and preprocessing module includes a data acquisition unit, a data preprocessing unit, and a feature extraction and selection unit. The output of the data acquisition unit is communicatively connected to the input of the data preprocessing unit, and the output of the data preprocessing unit is communicatively connected to the input of the feature extraction and selection unit. The acquisition unit is used to collect multi-dimensional performance data from the server; The data preprocessing unit is used to perform preprocessing on the data, such as standardization, missing value imputation, and outlier detection. The feature extraction and selection unit is used to automatically extract key features from the raw data using the autoencoder method in deep learning.