Virtual instance dynamic scheduling method based on AI cloud desktop

Abstract

The invention particularly relates to a virtual instance dynamic scheduling method based on an AI cloud desktop. According to the virtual instance dynamic scheduling method based on the AI cloud desktop, virtual instance data are collected from the cloud desktop in real time, and feature processing and feature extraction are carried out; introducing an LSTM neural network of an attention mechanism to construct a prediction model, after deep learning and training, inputting preprocessed real-time feature data, outputting and predicting a future system state and a user demand, and formulating and executing a corresponding scheduling strategy; and an execution result is fed back to the model, and scheduling algorithms in the prediction model and the intelligent scheduling module are continuously optimized. According to the virtual instance dynamic scheduling method based on the AI cloud desktop, by means of an artificial intelligence technology and a deep learning algorithm, resource waste is reduced, resource use peak pressure is relieved, the problems of system jamming, slow response and the like caused by insufficient resources in peak hours can be avoided, and the service life of the system is prolonged. And the resource utilization rate of the cloud desktop under multiple users and multiple scenes is effectively improved.

Description

Technical Field

[0001] This invention relates to the field of cloud computer terminal technology, and in particular to a method for dynamically scheduling virtualized instances based on AI cloud desktops. Background Technology

[0002] With the deep integration of cloud computing and artificial intelligence technologies, cloud desktop technology has made significant breakthroughs in performance, security, and intelligence, and its application scenarios are constantly expanding, gradually extending from the initial enterprise office field to multiple industries such as education, healthcare, finance, and design.

[0003] However, traditional cloud desktop virtualization instance scheduling often adopts static allocation or scheduling mode based on fixed rules.

[0004] In static allocation mode, when a virtualization instance is created, the system pre-allocates a fixed amount of CPU, memory, storage, and other resources to each instance, and these resource allocations remain unchanged throughout the entire usage cycle. While this approach is simple to implement, it suffers from a serious problem of resource waste.

[0005] While rule-based scheduling offers slightly more flexibility than static allocation, it typically relies on pre-defined, simple rules such as first-come, first-served or shortest-job-first-served principles for resource allocation and scheduling. In real-world multi-user, multi-scenario environments, user behavior and application resource consumption exhibit complex, dynamic changes that these fixed rules cannot adapt to in real-time and accurately.

[0006] To better schedule cloud desktop virtualization instances in different scenarios, cloud desktop virtualization instance scheduling integrates advanced technologies such as machine learning and big data analytics. This enables the analysis and processing of data on user behavior, application resource consumption, and system load, allowing for real-time and dynamic resource allocation based on predictions. This alleviates peak resource usage pressure and reduces user wait times. Therefore, this paper presents an AI-based dynamic scheduling method for cloud desktop virtualization instances, which is significant for real-time dynamic scheduling of virtualization instances across multiple users and scenarios.

[0007] To address the issues of resource waste and inability to dynamically allocate resources in traditional cloud desktop virtualization instance scheduling, this invention proposes a dynamic scheduling method for virtualization instances based on AI cloud desktops. Summary of the Invention

[0008] To overcome the shortcomings of existing technologies, this invention provides a simple and efficient method for dynamic scheduling of virtualized instances based on AI cloud desktops.

[0009] This invention is achieved through the following technical solution: A method for dynamically scheduling virtualized instances based on AI cloud desktops includes the following steps: Step S1: Data Acquisition and Preprocessing Collect virtualization instance data from cloud desktops in real time, including but not limited to user operation behavior, application resource consumption and system load data, and perform feature processing and feature extraction on the collected data; In step S1, the collected raw data is cleaned to remove noise and outliers, and then the data is preprocessed, including data normalization and feature extraction, to convert the raw data into a format suitable for deep learning algorithms.

[0010] In step S1, the TSNE algorithm or the MinmaxScaler tool is used to perform data normalization and feature extraction.

[0011] Step S2, Data Analysis and Scheduling An LSTM neural network with an attention mechanism is introduced to build a prediction model for feature data, and the prediction model is then subjected to in-depth learning and training. The preprocessed real-time feature data is input into the trained prediction model. Based on the input feature data, the model outputs predictions of future system state and user needs, providing a more accurate basis for dynamic scheduling. In step S2, the feature data obtained after preprocessing the original data is custom-divided into a training set, a validation set, and a test set. The prediction model is trained using the training set, and the model performance is optimized by adjusting the model's hyperparameters. During training, the prediction model processes the given time series data, captures long-term dependencies in the data, and predicts future load conditions. The validation set is used to evaluate the prediction model during training to prevent overfitting; finally, the test set is used to evaluate the trained prediction model to ensure that the model has good generalization ability.

[0012] Step S3: Performance Report Generation Based on the predicted system status and user needs, and using the trained prediction model, a corresponding scheduling strategy is formulated and executed to achieve dynamic optimization scheduling of cloud desktop virtualization instances. In step S3, when executing the scheduling strategy, the scheduling timing should be determined based on changes in system status and fluctuations in user demand. When the system load reaches a custom threshold, user demand changes abruptly, or resource utilization becomes abnormal, the scheduling operation should be triggered in a timely manner.

[0013] Step S4: Result Feedback and Optimization The execution results are fed back into the model to continuously optimize the prediction model and the scheduling algorithm in the intelligent scheduling module.

[0014] In step S4, various performance indicators of the cloud desktop system are collected and the data is sent to the monitoring server. The monitoring server stores and analyzes the monitoring data, generates a performance report, and feeds the performance report information back to the prediction model. The prediction model in the data processing and analysis module and the scheduling algorithm in the intelligent scheduling module are continuously optimized to improve the accuracy of the model's prediction of the cloud desktop load.

[0015] In step S4, the network parameters are continuously adjusted through the backpropagation algorithm to minimize the prediction error; the loss function used during training is the mean squared error loss function.

[0016] A multi-type peripheral adapter for an Android cloud computer terminal, comprising: The data acquisition and preprocessing module is responsible for collecting virtualization instance data from the cloud desktop in real time, including but not limited to user operation behavior, application resource consumption and system load data, and performing feature processing and feature extraction on the collected data. The data processing and analysis module is responsible for introducing an attention-based LSTM neural network, building a prediction model for feature data, and performing in-depth learning and model training on the prediction model. The preprocessed real-time feature data is input into the trained prediction model. Based on the input feature data, the model outputs predictions of future system state and user needs, providing a more accurate basis for dynamic scheduling. The intelligent scheduling module is responsible for formulating and executing corresponding scheduling strategies based on the predicted system status and user needs, using a trained prediction model, to achieve dynamic optimization scheduling of cloud desktop virtualization instances. The monitoring server is responsible for collecting various performance indicators of the cloud desktop system, storing and analyzing the monitoring data, generating performance reports, and feeding the performance report information back into the prediction model. It continuously optimizes the prediction model in the data processing and analysis module and the scheduling algorithm in the intelligent scheduling module to improve the accuracy of the model's prediction of cloud desktop load.

[0017] A computing device based on a dynamic scheduling method for virtualized instances using an AI cloud desktop includes: One or more processors, one or more memories, and one or more programs, wherein the one or more programs are stored in the one or more memories and configured to be executed by the one or more processors, and the one or more programs include instructions for performing any of the methods described above.

[0018] A computer-readable storage medium storing one or more programs, the one or more programs including instructions that, when executed by a computing device using a method for dynamically scheduling virtualized instances based on an AI cloud desktop, cause the computing device to perform any of the methods described above.

[0019] The beneficial effects of this invention are: the AI ​​cloud desktop-based virtualization instance dynamic scheduling method, with the help of artificial intelligence technology and deep learning algorithms, reduces resource waste and alleviates peak resource usage pressure, and can avoid system lag and slow response caused by insufficient resources during peak periods, effectively improving the resource utilization rate of cloud desktops in multi-user and multi-scenario environments. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Appendix Figure 1 This is a schematic diagram of the AI ​​cloud desktop-based virtualization instance dynamic scheduling method of the present invention. Detailed Implementation

[0022] To enable those skilled in the art to better understand the technical solutions of this invention, the technical solutions in the embodiments of this invention will be clearly and completely described below in conjunction with the embodiments of this invention. Obviously, the described embodiments are merely some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this invention.

[0023] This AI-based cloud desktop-based method for dynamically scheduling virtualized instances includes the following steps: Step S1: Data Acquisition and Preprocessing Collect virtualization instance data from cloud desktops in real time, including but not limited to user operation behavior, application resource consumption and system load data, and perform feature processing and feature extraction on the collected data; In step S1, the collected raw data is cleaned to remove noise and outliers. Then, the data is preprocessed, including data normalization and feature extraction, to convert the raw data into a format suitable for deep learning algorithms, ensuring data quality and providing a reliable data foundation for subsequent analysis.

[0024] In step S1, the TSNE algorithm or the MinmaxScaler tool is used to perform data normalization and feature extraction.

[0025] Step S2, Data Analysis and Scheduling The resource usage and user demand of cloud desktop systems change over time, exhibiting time-series characteristics. LSTM neural networks can process time-series data, capture long-term dependencies in the data, and adding an attention mechanism can further improve the model's focus on important features, thereby enhancing the model's prediction accuracy.

[0026] An LSTM neural network with an attention mechanism is introduced to build a prediction model for feature data, and the prediction model is then subjected to in-depth learning and training. The preprocessed real-time feature data is input into the trained prediction model. Based on the input feature data, the model outputs predictions of future system state and user needs, providing a more accurate basis for dynamic scheduling. In step S2, the feature data obtained after preprocessing the original data is custom-divided into a training set, a validation set, and a test set. The prediction model is trained using the training set, and the model performance is optimized by adjusting the model's hyperparameters. During training, the prediction model processes the given time series data, captures long-term dependencies in the data, predicts future load conditions, and helps the scheduling algorithm prepare for resource allocation in advance to avoid resource shortages or waste. The validation set is used to evaluate the prediction model during training to prevent overfitting; finally, the test set is used to evaluate the trained prediction model to ensure that the model has good generalization ability.

[0027] Step S3: Performance Report Generation Based on the predicted system status and user needs, and using the trained prediction model, a corresponding scheduling strategy is formulated and executed to achieve dynamic optimization scheduling of cloud desktop virtualization instances. In step S3, when executing the scheduling strategy, the timing and method of scheduling need to be considered. The scheduling timing should be determined based on changes in system state and fluctuations in user demand. When the system load reaches a custom threshold, user demand changes abruptly, or resource utilization becomes abnormal, the scheduling operation should be triggered promptly.

[0028] Step S4: Result Feedback and Optimization The execution results are fed back into the model to continuously optimize the prediction model and the scheduling algorithm in the intelligent scheduling module.

[0029] In step S4, various performance indicators of the cloud desktop system are collected and the data is sent to the monitoring server. The monitoring server stores and analyzes the monitoring data, generates a performance report, and feeds the performance report information back to the prediction model. The prediction model in the data processing and analysis module and the scheduling algorithm in the intelligent scheduling module are continuously optimized to improve the accuracy of the model's prediction of the cloud desktop load.

[0030] In step S4, the network parameters are continuously adjusted through the backpropagation algorithm to minimize the prediction error; the loss function used during training is the mean squared error loss function.

[0031] The Android cloud computer terminal supports various types of peripheral adapters, including: The data acquisition and preprocessing module is responsible for collecting virtualization instance data from the cloud desktop in real time, including but not limited to user operation behavior, application resource consumption and system load data, and performing feature processing and feature extraction on the collected data. The data processing and analysis module is responsible for introducing an attention-based LSTM neural network, building a prediction model for feature data, and performing in-depth learning and model training on the prediction model. The preprocessed real-time feature data is input into the trained prediction model. Based on the input feature data, the model outputs predictions of future system state and user needs, providing a more accurate basis for dynamic scheduling. The intelligent scheduling module is responsible for formulating and executing corresponding scheduling strategies based on the predicted system status and user needs, using a trained prediction model, to achieve dynamic optimization scheduling of cloud desktop virtualization instances. The monitoring server is responsible for collecting various performance indicators of the cloud desktop system, storing and analyzing the monitoring data, generating performance reports, and feeding the performance report information back into the prediction model. It continuously optimizes the prediction model in the data processing and analysis module and the scheduling algorithm in the intelligent scheduling module to improve the accuracy of the model's prediction of cloud desktop load.

[0032] This AI-based cloud desktop virtualization instance dynamic scheduling method computing device includes: One or more processors, one or more memories, and one or more programs, wherein the one or more programs are stored in the one or more memories and configured to be executed by the one or more processors, and the one or more programs include instructions for performing any of the methods described above.

[0033] The computer-readable storage medium stores one or more programs, the one or more programs including instructions that, when executed by a computing device using the AI ​​cloud desktop-based virtualization instance dynamic scheduling method, cause the computing device to perform any of the methods described above.

[0034] This AI-based cloud desktop virtualization instance dynamic scheduling method, with the help of artificial intelligence technology and deep learning algorithms, reduces resource waste and alleviates peak resource usage pressure. It can avoid system lag and slow response caused by insufficient resources during peak periods, and effectively improve the resource utilization rate of cloud desktops in multi-user and multi-scenario environments.

[0035] Compared with existing technologies, its main features are as follows: 1) Real-time data collection and processing ensures data timeliness; 2) By intelligently and dynamically scheduling cloud desktop virtualization instances, network response latency is reduced, enabling users to obtain computing results faster and improving response speed; 3) Virtualization instance scheduling based on real-time data and predictive models enables the rational allocation of computing resources, avoids resource waste and idleness, and improves resource utilization. 4) Predictive scheduling strategies can anticipate peak resource usage and allocate resources rationally to nodes that may experience high loads, thus alleviating peak resource usage pressure and ensuring stable system operation. 5) Through continuous learning and optimization, the adaptive ability and robustness of the prediction model have been improved; 6) Trigger the intelligent scheduling mechanism to realize operations such as autonomously calculating resource scheduling priorities and rationally allocating resources, thereby improving the user experience.

[0036] The embodiments described above are merely one specific implementation of the present invention. Ordinary changes and substitutions made by those skilled in the art within the scope of the technical solution of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for dynamically scheduling virtualized instances based on AI cloud desktops, characterized in that: Includes the following steps: Step S1: Data Acquisition and Preprocessing Collect virtualization instance data from cloud desktops in real time, including but not limited to user operation behavior, application resource consumption and system load data, and perform feature processing and feature extraction on the collected data; Step S2, Data Analysis and Scheduling An LSTM neural network with an attention mechanism is introduced to build a prediction model for feature data, and the prediction model is then subjected to in-depth learning and training. The preprocessed real-time feature data is input into the trained prediction model. Based on the input feature data, the model outputs predictions of future system state and user needs, providing a more accurate basis for dynamic scheduling. Step S3: Performance Report Generation Based on the predicted system status and user needs, and using the trained prediction model, a corresponding scheduling strategy is formulated and executed to achieve dynamic optimization scheduling of cloud desktop virtualization instances. Step S4: Result Feedback and Optimization The execution results are fed back into the model to continuously optimize the prediction model and the scheduling algorithm in the intelligent scheduling module.

2. The method for dynamic scheduling of virtualized instances based on AI cloud desktops according to claim 1, characterized in that: In step S1, the collected raw data is cleaned to remove noise and outliers, and then the data is preprocessed, including data normalization and feature extraction, to convert the raw data into a format suitable for deep learning algorithms.

3. The method for dynamic scheduling of virtualized instances based on AI cloud desktops according to claim 2, characterized in that: In step S1, the TSNE algorithm or the MinmaxScaler tool is used to perform data normalization and feature extraction.

4. The method for dynamic scheduling of virtualized instances based on AI cloud desktops according to claim 1, characterized in that: In step S2, the feature data obtained after preprocessing the original data is custom-divided into a training set, a validation set, and a test set. The prediction model is trained using the training set, and the model performance is optimized by adjusting the model's hyperparameters. During training, the prediction model processes the given time series data, captures long-term dependencies in the data, and predicts future load conditions. The validation set is used to evaluate the prediction model during training to prevent overfitting; finally, the test set is used to evaluate the trained prediction model to ensure that the model has good generalization ability.

5. The method for dynamic scheduling of virtualized instances based on AI cloud desktops according to claim 1, characterized in that: In step S3, when executing the scheduling strategy, the scheduling timing should be determined based on changes in system status and fluctuations in user demand. When the system load reaches a custom threshold, user demand changes abruptly, or resource utilization becomes abnormal, the scheduling operation should be triggered in a timely manner. In step S4, various performance indicators of the cloud desktop system are collected and the data is sent to the monitoring server. The monitoring server stores and analyzes monitoring data, generates performance reports, and feeds the performance report information back into the prediction model. It continuously optimizes the prediction model in the data processing and analysis module and the scheduling algorithm in the intelligent scheduling module to improve the accuracy of the model's prediction of cloud desktop load.

6. The method for dynamic scheduling of virtualized instances based on AI cloud desktops according to claim 5, characterized in that: In step S4, the network parameters are continuously adjusted through the backpropagation algorithm to minimize the prediction error; the loss function used during training is the mean squared error loss function.

7. A method for adapting multiple types of peripherals to an Android cloud computer terminal, characterized in that: To implement the method according to any one of claims 1 to 6, comprising: The data acquisition and preprocessing module is responsible for collecting virtualization instance data from the cloud desktop in real time, including but not limited to user operation behavior, application resource consumption and system load data, and performing feature processing and feature extraction on the collected data. The data processing and analysis module is responsible for introducing an attention-based LSTM neural network, building a prediction model for feature data, and performing in-depth learning and model training on the prediction model. The preprocessed real-time feature data is input into the trained prediction model. Based on the input feature data, the model outputs predictions of future system state and user needs, providing a more accurate basis for dynamic scheduling. The intelligent scheduling module is responsible for formulating and executing corresponding scheduling strategies based on the predicted system status and user needs, using a trained prediction model, to achieve dynamic optimization scheduling of cloud desktop virtualization instances. The monitoring server is responsible for collecting various performance indicators of the cloud desktop system, storing and analyzing the monitoring data, generating performance reports, and feeding the performance report information back into the prediction model. It continuously optimizes the prediction model in the data processing and analysis module and the scheduling algorithm in the intelligent scheduling module to improve the accuracy of the model's prediction of cloud desktop load.

8. A computing device with a dynamic scheduling method for virtualized instances based on AI cloud desktops, characterized in that: include: One or more processors, one or more memories, and one or more programs, wherein the one or more programs are stored in the one or more memories and configured to be executed by the one or more processors, the one or more programs including instructions for performing the method according to any one of claims 1 to 6.

9. A computer-readable storage medium for storing one or more programs, characterized in that: The one or more programs include instructions that, when executed by a computing device using the AI ​​cloud desktop-based virtualization instance dynamic scheduling method, cause the computing device to perform the method according to any one of claims 1 to 6.

Images