Cloud desktop scheduling method and system

Abstract

The invention relates to the technical field of cloud desktop management, and particularly discloses a cloud desktop scheduling method and system, and the method comprises the steps: collecting user behavior data, the real usage amount of various resources, and system performance data; processing the user behavior data to obtain the cloud desktop user viscosity, processing the system performance data to obtain a cloud desktop system performance evaluation index, combining the real usage amount of various resources, and performing comprehensive analysis to obtain the predicted usage amount of various resources at each time node; according to the predicted usage amount of each type of resources and the real usage amount of each type of resources of each historical time node, a resource usage amount prediction precision index is obtained through comprehensive analysis; and evaluating a resource usage amount prediction result according to the resource usage amount prediction precision index. According to the method, the predicted usage amount of the future time node is evaluated, so that the stability and the safety of the system are improved, the resource use can be optimized, the user experience is improved, and higher operation efficiency is brought to an enterprise.

Description

Technical Field

[0001] The present invention relates to the technical field of cloud desktop management, and specifically to a cloud desktop scheduling method and system. Background Art

[0002] Cloud desktop scheduling is a very important part in the field of cloud computing. With the development of artificial intelligence technology and cloud computing technology, providing more user-friendly cloud desktop services has become the norm. Efficient and user-friendly cloud desktop scheduling methods are of great significance to improving user experience and enhancing market competitiveness.

[0003] For example, the invention patent with announcement number CN108769233B is a resource optimization allocation method based on desktop cloud, including a resource optimization allocation method based on desktop cloud, the following steps: S1: construct a laboratory resource optimization model based on professional teaching desktop cloud D based on professional teaching, teachers T defined by the scope of teaching courses, and students S who complete the semester course goals; calculate the ideal number of desktop clouds, recalculate the minimum number of desktop clouds based on the limit on the number of professional teachers, and try to fill the physical laboratories where experimental teaching is arranged as much as possible based on the determined number of desktop clouds D.

[0004] For example, the invention patent with announcement number CN104253865B is a two-level management method for a hybrid desktop cloud service platform, which includes the following steps: building a hybrid desktop cloud data center; the server node sends a registration request to the central management node (Vmm-Server); the central management node receives and processes the registration request of the server node; the central management node uniformly receives user requests through the service interface layer; the central management node responds to user requests according to a two-layer scheduling method; the central management node uses a two-layer migration method to maintain system load balancing according to the current system status and virtual machine operation status.

[0005] However, in the process of implementing the technical solution of the invention in the embodiments of the present application, it was found that the above technology has at least the following technical problems: the current cloud desktop scheduling method focuses more on improving resource utilization, but resources may still be unevenly distributed during peak periods, and users' personalized needs cannot be fully met. Summary of the invention

[0006] In view of the deficiencies in the prior art, the present invention application provides a cloud desktop scheduling method and system, which can effectively solve the problems involved in the above-mentioned background technology.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: The first aspect of the present invention provides a cloud desktop scheduling method, including: collecting user behavior data, actual usage of various resources and system performance data.

[0008] The user behavior data is processed to obtain the cloud desktop user stickiness, and the system performance data is processed to obtain the cloud desktop system performance evaluation index. Combined with the actual usage of various resources, a comprehensive analysis is performed to obtain the predicted usage of various resources at each time node.

[0009] The time nodes are divided according to the current time node to obtain historical time nodes and future time nodes, and the resource usage prediction accuracy index is obtained through comprehensive analysis based on the predicted usage of various resources and the actual usage of various resources at each historical time node.

[0010] The resource usage prediction results are evaluated according to the resource usage prediction accuracy index. If the evaluation result is qualified, cloud desktop resource scheduling is performed according to the various predicted usages at each future time node. If the evaluation result is unqualified, the resource usage prediction is performed again.

[0011] As a further method, the user behavior data is processed to obtain the cloud desktop user stickiness, and the specific processing process is: the user behavior data includes the cumulative number of logins, the cumulative login time and the number of users.

[0012] The cloud desktop user viscosity evaluation influencing factor corresponding to the preset cumulative number of logins and the cloud desktop user viscosity evaluation influencing factor corresponding to the cumulative login duration are extracted from the cloud desktop database.

[0013] Based on the cumulative number of logins and cumulative login duration of each user at each time node in each monitoring cycle, a comprehensive analysis is performed to obtain the cloud desktop user stickiness at each time node in each monitoring cycle.

[0014] As a further method, the system performance data is processed to obtain a cloud desktop system performance evaluation index, and the specific processing process is: the system performance data includes system response time and system throughput.

[0015] A cloud desktop system performance evaluation influencing factor corresponding to a preset system response time and a cloud desktop system performance evaluation influencing factor corresponding to a preset system throughput are extracted from the cloud desktop database.

[0016] Based on the system response time and system throughput at each time node in each monitoring cycle, a comprehensive analysis is performed to obtain the cloud desktop system performance evaluation index.

[0017] As a further method, the comprehensive analysis obtains the predicted usage of various resources at each time node. The specific analysis process is: extracting the preset cloud desktop user stickiness, cloud desktop system performance evaluation index and predicted usage evaluation influencing factors corresponding to the actual usage of various resources from the cloud desktop database.

[0018] Based on user stickiness, system performance evaluation index and the actual usage of various resources, a comprehensive analysis is conducted to obtain the predicted usage of various resources at each time node.

[0019] As a further method, the resource usage prediction accuracy index is a quantitative indicator obtained by comparing and analyzing the predicted usage and actual usage of each resource at each time node, and is used to quantify the accuracy of resource usage prediction.

[0020] As a further method, the cloud desktop system performance evaluation index is a quantitative indicator obtained by analyzing the system response time and system throughput of each time node in each monitoring cycle, and is used to quantify the performance of the cloud desktop system.

[0021] As a further method, the cloud desktop scheduling method is evaluated according to the resource usage prediction accuracy index. The specific evaluation process is: extracting the resource usage prediction accuracy threshold in the cloud desktop database, comparing the resource usage prediction accuracy index with the resource usage prediction accuracy threshold, if the resource usage prediction accuracy index is greater than or equal to the resource usage prediction accuracy threshold, then the cloud desktop scheduling method is evaluated as qualified; if the resource usage prediction accuracy index is less than the resource usage prediction accuracy threshold, then the cloud desktop scheduling method is evaluated as unqualified.

[0022] As a further method, the predicted usage of each resource at each time node is specifically expressed as:

[0023]

[0024] Among them, y ik represents the predicted usage of the kth resource at the i-th time node, x ijk represents the usage of the kth resource at the i-th time node on the j-th day, σ ik represents the resource prediction compensation parameter of the kth resource at the ith time node, ω1 represents the prediction usage evaluation impact factor corresponding to the set historical average usage, ω2 represents the prediction usage evaluation impact factor corresponding to the set resource prediction compensation parameter, i represents the number of time nodes, i=1,2,3,...,n, n represents the total number of time nodes, j represents the number of monitoring cycles, j=1,2,3,...,m, m represents the total number of monitoring cycles, k represents the number of resources, k=1,2,3,...,h, h represents the total number of resources.

[0025] As a further method, the prediction accuracy index, the specific numerical expression is:

[0026]

[0027] Among them, C represents the prediction accuracy index, e represents the natural constant, and yik represents the predicted usage of the kth resource at the i-th time node, represents the actual usage of the kth resource at the i-th time node, Δy represents the set allowable deviation usage, i represents the number of time nodes, i=1,2,3,...,n, n represents the total number of time nodes, k represents the number of resources, k=1,2,3,...,h, h represents the total number of resources.

[0028] The second aspect of the present application provides a cloud desktop scheduling method system, including: a cloud desktop data collection module, used to collect user behavior data, actual usage of various resources and system performance data, the user behavior data includes the user's cumulative login times, cumulative login duration and system response time; the system performance data includes system throughput and resource usage.

[0029] The cloud desktop predicted usage analysis module is used to process user behavior data to obtain the cloud desktop user stickiness, and to process system performance data to obtain the cloud desktop system performance evaluation index. Combined with the actual usage of various resources, a comprehensive analysis is performed to obtain the predicted usage of various resources at each time node.

[0030] The cloud desktop prediction accuracy analysis module is used to divide the time nodes according to the current time node, obtain each historical time node and each future time node, and obtain the resource usage prediction accuracy index based on the predicted usage of each type of resources and the actual usage of each type of resources at each historical time node through comprehensive analysis.

[0031] The cloud desktop scheduling method evaluation module is used to evaluate the resource usage prediction results according to the resource usage prediction accuracy index. If the evaluation result is qualified, the cloud desktop resource scheduling is performed according to the various predicted usages at each future time node. If the evaluation result is unqualified, the resource usage prediction is performed again.

[0032] Cloud desktop database, cloud desktop data includes cloud desktop user stickiness evaluation influencing factors corresponding to preset cumulative login times and cloud desktop user stickiness evaluation influencing factors corresponding to cumulative login duration, cloud desktop system performance evaluation influencing factors corresponding to preset system response time and cloud desktop system performance evaluation influencing factors corresponding to preset system throughput, resource usage and predicted usage evaluation influencing factors corresponding to resource prediction compensation parameters, and resource usage prediction accuracy threshold.

[0033] Compared with the prior art, the embodiments of the present invention have at least the following advantages or beneficial effects:

[0034] (1) The present invention can help formulate more targeted marketing strategies and improve user retention and satisfaction by evaluating the stickiness of cloud desktop users. Secondly, abnormal changes in user stickiness may indicate problems in the service, which helps to quickly diagnose and solve the problems. At the same time, it can more accurately predict resource requirements and avoid cost waste caused by over-configuration.

[0035] (2) The present invention can more accurately predict resource requirements and avoid cost waste caused by over-configuration by evaluating the performance of the cloud desktop system. At the same time, according to the changes in the cloud desktop system performance evaluation index, more personalized services and support can be provided, such as providing additional support for high-load periods.

[0036] (3) The present invention not only helps to improve the stability and security of the system by evaluating the predicted usage at future time nodes, but also optimizes resource utilization, enhances user experience, and brings higher operational efficiency to the enterprise. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] The accompanying drawings are used to further illustrate the present invention, but the embodiments in the accompanying drawings do not constitute any limitation to the present invention. A person of ordinary skill in the art can obtain other drawings based on the following drawings without creative work.

[0038] Figure 1 This is a flow chart of the cloud desktop scheduling method applied for in the present invention.

[0039] Figure 2 This is a schematic diagram of the module connection of the cloud desktop system applied for in the present invention.

[0040] Figure 3 It is a schematic diagram of the functional relationship between the resource prediction correction parameter evaluation index and the cloud desktop user stickiness evaluation index involved in the present invention application. DETAILED DESCRIPTION

[0041] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0042] Reference Figure 1 As shown, the first aspect of the present invention application provides a cloud desktop scheduling method, comprising:

[0043] Collect user behavior data, actual usage of various resources, and system performance data;

[0044] The user behavior data is processed to obtain the cloud desktop user stickiness, and the system performance data is processed to obtain the cloud desktop system performance evaluation index. Combined with the actual usage of various resources, a comprehensive analysis is performed to obtain the predicted usage of various resources at each time node;

[0045] The time nodes are divided according to the current time node to obtain each historical time node and each future time node, and the resource usage prediction accuracy index is obtained through comprehensive analysis based on the predicted usage of each type of resource at each historical time node and the actual usage of each type of resource;

[0046] The resource usage prediction results are evaluated according to the resource usage prediction accuracy index. If the evaluation result is qualified, cloud desktop resource scheduling is performed according to the various predicted usages at each future time node. If the evaluation result is unqualified, the resource usage prediction is performed again.

[0047] It should be understood that the actual usage of various types of resources in this embodiment include: CPU usage, memory usage, system disk usage, data disk usage, bandwidth usage, traffic usage and GPU usage.

[0048] Specifically, the user behavior data is processed to obtain the cloud desktop user stickiness, and the specific processing process is: the user behavior data includes the cumulative number of logins, the cumulative login duration and the number of users.

[0049] The cloud desktop user stickiness evaluation influencing factor corresponding to the preset cumulative login times and the cloud desktop user stickiness evaluation influencing factor corresponding to the cumulative login duration can be extracted from the cloud desktop database.

[0050] Based on the cumulative number of logins and cumulative login duration of each user at each time node in each monitoring cycle, a comprehensive analysis is performed to obtain the cloud desktop user stickiness at each time node in each monitoring cycle.

[0051] In a specific embodiment, the functions provided by the cloud desktop platform are used to monitor the cumulative number of user logins, the cumulative login duration, and the number of users. By monitoring the cumulative number of logins and the number of users, it is possible to understand the load of the system in different time periods, thereby further allocating resources to achieve load balancing, and at the same time, it is helpful to predict future resource requirements, so as to make expansion plans in advance and ensure that the system can cope with the load during peak periods; monitoring the cumulative login duration helps to promptly discover performance anomalies and security vulnerabilities, so as to quickly locate and resolve problems. Abnormally high login times or login durations may also be signs of security threats.

[0052] Specifically, the cloud desktop user stickiness evaluation index has a specific numerical expression:

[0053]

[0054] Among them, A ij represents the cloud desktop user stickiness evaluation index at the i-th time node in the j-th monitoring period, T ijf represents the cumulative login times of the fth user at the i-th time node in the j-th monitoring cycle, T0 represents the critical cumulative login times, S ijf represents the cumulative login time of the fth user at the i-th time node in the j-th monitoring cycle, S0 represents the critical cumulative login time, f0 represents the critical number of users, Indicates the cloud desktop user stickiness evaluation impact factor corresponding to the set cumulative login times. Indicates the impact factor of cloud desktop user stickiness evaluation corresponding to the set cumulative login time. It represents the influencing factor of cloud desktop user stickiness evaluation corresponding to the set number of users, i represents the number of time nodes, i=1,2,3,...,n, n represents the total number of time nodes, j represents the number of monitoring cycles, j=1,2,3,...,m, m represents the total number of monitoring cycles, f represents the number of users, f=1,2,3,...,z, z represents the total number of users.

[0055] The algorithm of this embodiment combines the cumulative number of logins and the cumulative login duration of users at a certain time node and the number of users at the current time node, and comprehensively analyzes to obtain the dependent variable. In this formula, the more cumulative logins each user has, the longer the cumulative login duration is, indicating that the user is more dependent on the cloud desktop service. At the same time, the increase in the number of users indicates that the cloud desktop service is more popular. Comprehensive analysis can obtain a more comprehensive cloud desktop user stickiness evaluation index.

[0056] Table 1 Example of cloud desktop user stickiness evaluation index data

[0057]

[0058] As shown in Table 1, the cloud desktop user stickiness evaluation index is jointly determined by the cumulative number of logins, the cumulative login duration and the number of users. In a specific embodiment, z=1, the critical cumulative number of logins is 10 times, the critical cumulative login duration is 1 hour, the critical number of users is 50, and the cloud desktop user stickiness evaluation impact factor corresponding to the set cumulative number of logins is 0.3, the cloud desktop user stickiness evaluation impact factor corresponding to the set cumulative login duration is 0.4, and the cloud desktop user stickiness evaluation impact factor corresponding to the set number of users is 0.3.

[0059] This formula takes into account the cumulative number of logins, cumulative login time and number of users for each user, and can provide more personalized services and support based on the user's behavior pattern, such as providing higher-level technical support or customized services for high-frequency users. By standardizing the cumulative number of logins, cumulative login time and number of users at different time nodes in different monitoring periods, it ensures that they are compared at the same level, which improves the fairness and comparability of the evaluation. At the same time, it can understand which time periods or which user groups have a high degree of dependence on cloud desktop services, so as to reasonably arrange resource allocation and optimize user experience. By weighting the impact of the cumulative number of logins, cumulative login time and number of users, it reflects their relative importance in the evaluation index. The weights of different factors can be adjusted according to different needs, making the model highly adaptable. It can be seen that the greater the cumulative number of logins, cumulative login time or number of users, the greater the cloud desktop user stickiness evaluation index. By evaluating the stickiness of cloud desktop users, we can help develop more targeted marketing strategies and improve user retention and satisfaction. Secondly, abnormal changes in user stickiness may indicate problems in the service, which helps to quickly diagnose and solve problems. At the same time, it can more accurately predict resource requirements and avoid cost waste caused by over-configuration.

[0060] In a specific embodiment, the value range of the cloud desktop user stickiness evaluation influencing factor corresponding to the cumulative number of logins, cumulative login duration and number of users of each user is between 0 and 1. By adjusting the value of the influencing factor, the influence of different factors on the final cloud desktop user stickiness evaluation index can be flexibly adjusted.

[0061] It should be understood that in this embodiment, a mapping set of the cumulative number of logins, the cumulative login duration, the number of users and the corresponding cloud desktop user stickiness evaluation influencing factors is constructed through the relationship between the cumulative number of logins, the cumulative login duration, the number of users in the historical data and the cloud desktop user stickiness evaluation index. The real-time cumulative number of logins, the cumulative login duration and the number of users are input, and the corresponding cloud desktop user stickiness evaluation influencing factors are obtained from the mapping set.

[0062] In a specific embodiment, the cloud desktop user stickiness evaluation index is a quantitative indicator obtained by analyzing the cumulative number of logins, the cumulative login duration, and the number of users, and is used to quantify the stickiness of cloud desktop users.

[0063] Specifically, the system performance data is processed to obtain the cloud desktop system performance evaluation index, and the specific processing process is: the system performance data includes system response time and system throughput.

[0064] The cloud desktop system performance evaluation influencing factor corresponding to the preset system response time and the cloud desktop system performance evaluation influencing factor corresponding to the preset system throughput can be extracted from the cloud desktop database.

[0065] Based on the system response time and system throughput at each time node in each monitoring cycle, a comprehensive analysis is performed to obtain the cloud desktop system performance evaluation index.

[0066] In a specific embodiment, throughput refers to the number of requests processed by the system per unit time, and response time refers to the time it takes for the system to respond to a request, that is, the time it takes for the entire process from when a user initiates a request on the client to when the client receives a response returned from the server. The system response time and system throughput can be detected through the built-in performance monitoring function of the cloud desktop platform. By monitoring the system response time, it can be ensured that the cloud desktop can respond quickly to user operations and provide a smooth user experience, while reducing user waiting time and improving work efficiency and satisfaction. Monitoring system throughput can help determine the system load, achieve reasonable allocation of resources, and ensure that resources are available when needed. Abnormal response time and throughput may indicate that the system is under attack or has potential security vulnerabilities. Detection can help identify areas where security measures need to be strengthened.

[0067] Specifically, the cloud desktop system performance evaluation index, the specific numerical expression is:

[0068]

[0069] Among them, B ij represents the performance evaluation index of the cloud desktop system at the i-th time node in the j-th monitoring cycle, P ij represents the system response time of the i-th time node in the j-th monitoring cycle, P0 represents the set maximum allowed system response time, Q ij represents the system throughput at the i-th time node in the j-th monitoring cycle, Q0 represents the set maximum allowed system throughput, ρ1 represents the cloud desktop system performance evaluation impact factor corresponding to the set response time, ρ2 represents the cloud desktop system performance evaluation impact factor corresponding to the set throughput, i represents the number of time nodes, i=1,2,3,...,n, n represents the total number of time nodes, j represents the number of monitoring cycles, j=1,2,3,...,m, m represents the total number of monitoring cycles.

[0070] The algorithm of this embodiment combines the system response time and system throughput of each time node, and comprehensively analyzes to obtain the dependent variable. There is a direct correlation between the system response time and the system throughput. When the system throughput increases, the system may need to process more requests, which may cause the response time of a single request to become longer. Conversely, if the system response time is too long, the system may not be able to effectively process more requests, thereby limiting the throughput. The comprehensive analysis can evaluate the system performance level at a specific time node within a specific monitoring period, and obtain a more comprehensive cloud desktop system performance evaluation index.

[0071] It should be explained that two key factors, namely, system response time and system throughput, are considered in this embodiment, which can help identify performance bottlenecks, quickly locate the causes of performance problems, and identify periods of low resource utilization efficiency, thereby saving costs. By standardizing the system response time and system throughput of different time nodes, ensuring that they are compared at the same level, the fairness and comparability of the evaluation are improved, and at the same time, the performance of the cloud desktop service in different time periods can be understood, so as to better allocate resources and achieve load balancing. By weighting the impact of system response time and system throughput, their relative importance in the evaluation index can be reflected, and the weights of different factors can be adjusted according to different needs, so that the model has good adaptability. It is not difficult to see that when the system response time is smaller or the system throughput is larger, the cloud desktop system performance evaluation index is larger. By evaluating the performance of the cloud desktop system, resource requirements can be predicted more accurately, and cost waste caused by over-configuration can be avoided. At the same time, according to the changes in the cloud desktop system performance evaluation index, more personalized services and support can be provided, such as providing additional support for high-load periods.

[0072] In a specific embodiment, the value range of the cloud desktop system performance evaluation influencing factors corresponding to the system response time and system throughput of each time node is between 0 and 1. By adjusting the value of the influencing factor, the influence of different factors on the final cloud desktop system performance evaluation index can be flexibly adjusted.

[0073] It should be understood that in this embodiment, a mapping set of system response time and system throughput and corresponding cloud desktop system performance evaluation influencing factors is constructed through the relationship between system response time and system throughput in historical data and the cloud desktop system performance evaluation index, and the real-time system response time and system throughput are input to obtain the corresponding cloud desktop system performance evaluation influencing factors from the mapping set.

[0074] In a specific embodiment, the cloud desktop system performance evaluation index is a quantitative indicator obtained by analyzing the system response time and the system throughput, and is used to quantify the performance of the cloud desktop system.

[0075] Specifically, a comprehensive analysis is performed to obtain the predicted usage of various types of resources at each time node. The specific analysis process is: extracting the preset cloud desktop user stickiness, cloud desktop system performance evaluation index and the predicted usage evaluation influencing factors corresponding to the actual usage of various types of resources from the cloud desktop database.

[0076] Based on user stickiness, system performance evaluation index and the actual usage of various resources, a comprehensive analysis is conducted to obtain the predicted usage of various resources at each time node.

[0077] Specifically, the resource forecast correction parameter evaluation index at each time node is expressed as follows:

[0078]

[0079] Among them, ε i A represents the resource forecast correction parameter evaluation index at the i-th time node, ij represents the cloud desktop user stickiness evaluation index at the i-th time node in the j-th monitoring period, B ij represents the cloud desktop system performance evaluation index of the i-th time node in the j-th monitoring cycle, θ1 represents the resource prediction correction parameter evaluation impact factor corresponding to the set user stickiness, θ2 represents the resource prediction correction parameter evaluation impact factor corresponding to the set system performance evaluation index, i represents the number of time nodes, i=1,2,3,...,n, n represents the total number of time nodes, j represents the number of monitoring cycles, j=1,2,3,...,m, m represents the total number of monitoring cycles.

[0080] The algorithm of this embodiment combines the cloud desktop user stickiness evaluation index and the cloud desktop system performance evaluation index at a certain time node, and comprehensively analyzes to obtain the dependent variable. If the system performance is good, the user stickiness will usually be high, because good performance helps to improve the user experience, thereby increasing the user's dependence on the service, indicating that more resources are needed to maintain the service quality. Comprehensive analysis can more comprehensively evaluate resource requirements and adjust resource forecasts accordingly.

[0081] like Figure 3 As shown, in a specific embodiment, m = 1, θ1 = 0.5, θ2 = 0.3. ij =0.1, the functional relationship between the resource prediction correction parameter evaluation index and the cloud desktop user stickiness evaluation index is shown in curve a; when B ij =0.5, the functional relationship between the resource prediction correction parameter evaluation index and the cloud desktop user stickiness evaluation index is shown in curve b; when B ij When =1, the functional relationship between the resource prediction correction parameter evaluation index and the cloud desktop user stickiness evaluation index is shown in curve c.

[0082] It should be explained that in this embodiment, two key factors, namely, the cloud desktop user viscosity evaluation index and the cloud desktop system performance evaluation index, are comprehensively considered, which can more accurately predict future resource requirements. According to the changing trends of these indicators, resource allocation can be dynamically adjusted to ensure that resources are available when needed. By summing and averaging the cloud desktop user viscosity and cloud desktop system performance evaluation index of different monitoring cycles, it is beneficial to accurately allocate resource requirements at future time nodes according to the cloud desktop user viscosity and cloud desktop system performance evaluation index at different time nodes, achieve load balancing, and improve user experience. By weighting the impact of cloud desktop user viscosity and cloud desktop system performance, it reflects their relative importance in the evaluation index. The weights of different factors can be adjusted according to different needs, so that the model has good adaptability. When the cloud desktop user viscosity evaluation index is larger and or the cloud desktop system performance evaluation index is larger, the resource prediction correction parameter evaluation index is larger. By evaluating the resource prediction correction parameter, the degree of influence of the cloud desktop user viscosity evaluation index and the cloud desktop system performance evaluation index on the resource prediction usage can be quantified, which helps to help management make more informed decisions, such as increasing or reducing resources in a timely manner.

[0083] In a specific embodiment, the value range of the resource prediction correction parameter evaluation influencing factor corresponding to the cloud desktop user stickiness evaluation index and the cloud desktop system performance evaluation index at each time node is between 0 and 1. By adjusting the value of the influencing factor, the influence of different factors on the final resource prediction correction parameter evaluation index can be flexibly adjusted.

[0084] It should be understood that in this embodiment, a mapping set of the cloud desktop user stickiness evaluation index and the cloud desktop system performance evaluation index and the corresponding resource prediction correction parameter evaluation influencing factors is constructed through the relationship between the cloud desktop user stickiness evaluation index and the cloud desktop system performance evaluation index in the historical data and the resource prediction correction parameter evaluation index, and the real-time cloud desktop user stickiness evaluation index and the cloud desktop system performance evaluation index are input to obtain the corresponding resource prediction correction parameter evaluation influencing factors from the mapping set.

[0085] In a specific embodiment, the resource prediction correction parameter evaluation index is a quantitative index obtained by analyzing the cloud desktop user stickiness evaluation index and the cloud desktop system performance evaluation index, and is used to quantify the resource prediction correction parameter.

[0086] Specifically, the predicted usage of each resource at each time node is expressed as follows:

[0087]

[0088] Among them, y ik represents the predicted usage of the kth resource at the i-th time node, xijk represents the usage of the kth resource at the i-th time node on the j-th day, σ ik represents the resource prediction compensation parameter of the kth resource at the ith time node, ω1 represents the prediction usage evaluation impact factor corresponding to the set historical average usage, ω2 represents the prediction usage evaluation impact factor corresponding to the set resource prediction compensation parameter, i represents the number of time nodes, i=1,2,3,...,n, n represents the total number of time nodes, j represents the number of monitoring cycles, j=1,2,3,...,m, m represents the total number of monitoring cycles, k represents the number of resources, k=1,2,3,...,h, h represents the total number of resources.

[0089] The algorithm of this embodiment combines the resource prediction usage and resource prediction compensation parameters of historical time nodes, and comprehensively analyzes to obtain the dependent variable. If there is a large deviation between the historical average usage and the actual usage, the resource prediction compensation parameter can be used to adjust the predicted value to make it closer to the actual usage to cope with uncertainty and changes. The comprehensive analysis can more accurately predict resource usage and perform resource scheduling accordingly.

[0090] It should be explained that in this embodiment, the resource prediction compensation parameters corresponding to the resource prediction correction parameter evaluation index at different time nodes are extracted from the cloud desktop data set, and the resource prediction compensation parameters can ensure that they are at the same order of magnitude as the resource usage.

[0091] It should be explained that in this embodiment, two key factors, namely, the resource usage of historical time nodes and the resource prediction compensation parameter, are comprehensively considered, and the demand for each time node and each resource can be predicted more accurately. By summing and averaging the resource usage of historical time nodes in different monitoring cycles, the user's behavior pattern can be further understood, and more personalized services can be provided accordingly. By weighting the influence of the resource usage of historical time nodes and the resource prediction compensation parameter, their relative importance in the evaluation index can be reflected, and the weights of different factors can be adjusted according to different needs, so that the model has good adaptability. It is not difficult to see that the greater the resource usage of the historical time node and or the greater the resource prediction compensation parameter, the greater the predicted usage of the future time node. By evaluating the predicted usage of the future time node, the influence of the resource usage of the historical time node and the resource prediction compensation parameter on the predicted usage of the future time node can be quantified, which not only helps to improve the stability and security of the system, but also optimizes resource usage, improves user experience, and brings higher operational efficiency to the enterprise.

[0092] In a specific embodiment, the value range of the influencing factor of the future time node resource prediction usage corresponding to the historical time node resource usage and the resource prediction compensation parameter is between 0 and 1. By adjusting the value of the influencing factor, the influence of different factors on the final future time node resource prediction usage can be flexibly adjusted.

[0093] It should be understood that in this embodiment, a mapping set of historical time node resource usage and resource prediction compensation parameters and corresponding future time node resource prediction usage influencing factors is constructed through the relationship between historical time node resource usage and resource prediction compensation parameters in historical data and future time nodes resource prediction usage, and the real-time historical time node resource usage and resource prediction compensation parameters are input to obtain the corresponding future time node resource prediction usage influencing factors from the mapping set.

[0094] Specifically, the resource usage prediction accuracy index is a quantitative indicator obtained by comparing and analyzing the predicted usage and actual usage of each resource at each time node, and is used to quantify the accuracy of resource usage prediction.

[0095] Furthermore, the resource usage prediction accuracy index has a specific numerical expression as follows:

[0096]

[0097] Among them, C represents the prediction accuracy index, e represents the natural constant, and y ik represents the predicted usage of the kth resource at the i-th time node, represents the actual usage of the kth resource at the i-th time node, Δy represents the set allowable deviation usage, i represents the number of time nodes, i=1,2,3,...,n, n represents the total number of time nodes, k represents the number of resources, k=1,2,3,...,h, h represents the total number of resources.

[0098] It should be explained that in this embodiment, it is not difficult to see that the closer the predicted usage of each resource at each time node is to the actual usage, the greater the resource usage prediction accuracy index. By evaluating the resource usage prediction accuracy index, the accuracy of the cloud desktop resource usage prediction can be effectively improved, thereby providing better services for enterprises and individual users, while reducing costs and improving efficiency.

[0099] Specifically, the cloud desktop scheduling method is evaluated according to the resource usage prediction accuracy index. The specific evaluation process is: extracting the resource usage prediction accuracy threshold in the cloud desktop database, comparing the resource usage prediction accuracy index with the resource usage prediction accuracy threshold, if the resource usage prediction accuracy index is greater than or equal to the resource usage prediction accuracy threshold, then the cloud desktop scheduling method is evaluated as qualified; if the resource usage prediction accuracy index is less than the resource usage prediction accuracy threshold, then the cloud desktop scheduling method is evaluated as unqualified.

[0100] Reference Figure 2 As shown, the second aspect of the present invention application provides a cloud desktop scheduling system, including:

[0101] The cloud desktop data collection module is used to collect user behavior data, actual usage of various resources and system performance data. User behavior data includes the cumulative number of user logins, cumulative login duration and system response time; system performance data includes system throughput and resource usage;

[0102] The cloud desktop predicted usage analysis module is used to process user behavior data to obtain the cloud desktop user stickiness, and process system performance data to obtain the cloud desktop system performance evaluation index. Combined with the actual usage of various resources, a comprehensive analysis is performed to obtain the predicted usage of various resources at each time node;

[0103] The cloud desktop prediction accuracy analysis module is used to divide the time nodes according to the current time node, obtain each historical time node and each future time node, and obtain the resource usage prediction accuracy index based on the predicted usage of each type of resources and the actual usage of each type of resources at each historical time node through comprehensive analysis;

[0104] The cloud desktop scheduling method evaluation module is used to evaluate the resource usage prediction results according to the resource usage prediction accuracy index. If the evaluation result is qualified, the cloud desktop resource scheduling is performed according to the various predicted usages at each future time node. If the evaluation result is unqualified, the resource usage prediction is performed again.

[0105] In a specific embodiment, when the evaluation result of the resource usage prediction result is qualified, cloud desktop resource scheduling is performed according to the various predicted usages of each future time node. For example, through analysis of historical data, it is found that 9:00 to 11:00 a.m. every day is the peak period for user logins, and the CPU and memory usage rates increase significantly at this time. Then, based on the prediction results, resource allocation is adjusted in advance, such as increasing the number of CPUs and memory capacity to cope with peak demand.

[0106] The cloud desktop database is used to store cloud desktop data. The cloud desktop data includes the cloud desktop user viscosity evaluation influencing factors corresponding to the preset cumulative number of logins and the cloud desktop user viscosity evaluation influencing factors corresponding to the cumulative login duration, the cloud desktop system performance evaluation influencing factors corresponding to the preset system response time and the cloud desktop system performance evaluation influencing factors corresponding to the preset system throughput, the resource usage and the predicted usage evaluation influencing factors corresponding to the resource prediction compensation parameters, and the resource usage prediction accuracy threshold. The data in the cloud desktop database can also be obtained by collecting data from multiple cloud desktop experiments.

[0107] The above contents are merely examples and explanations of the structure of the present invention. The technical personnel in this technical field may make various modifications or additions to the specific embodiments described or replace them in a similar manner. As long as they do not deviate from the structure of the invention or exceed the scope defined in the present invention, they should all fall within the protection scope of the present invention.

Claims

1. A cloud desktop scheduling method, characterized in that: include: Collect user behavior data, actual usage of various resources, and system performance data; The user behavior data is processed to obtain the cloud desktop user stickiness, and the system performance data is processed to obtain the cloud desktop system performance evaluation index. Combined with the actual usage of various resources, a comprehensive analysis is performed to obtain the predicted usage of various resources at each time node; The time nodes are divided according to the current time node to obtain each historical time node and each future time node, and the resource usage prediction accuracy index is obtained through comprehensive analysis based on the predicted usage of each type of resource at each historical time node and the actual usage of each type of resource; The resource usage prediction results are evaluated according to the resource usage prediction accuracy index. If the evaluation result is qualified, cloud desktop resource scheduling is performed according to the various predicted usages at each future time node. If the evaluation result is unqualified, the resource usage prediction is performed again.

2. The cloud desktop scheduling method according to claim 1, characterized in that: The user behavior data is processed to obtain the cloud desktop user stickiness, and the specific processing process is: The user behavior data includes the cumulative number of logins, the cumulative login duration and the number of users; Extracting a preset cloud desktop user stickiness evaluation influencing factor corresponding to the cumulative number of logins and a cloud desktop user stickiness evaluation influencing factor corresponding to the cumulative login duration from the cloud desktop database; Based on the cumulative number of logins and cumulative login duration of each user at each time node in each monitoring cycle, a comprehensive analysis is performed to obtain the cloud desktop user stickiness at each time node in each monitoring cycle.

3. The cloud desktop scheduling method according to claim 1, characterized in that: The system performance data is processed to obtain the cloud desktop system performance evaluation index, and the specific processing process is as follows: The system performance data includes system response time and system throughput; Extracting a cloud desktop system performance evaluation influencing factor corresponding to a preset system response time and a cloud desktop system performance evaluation influencing factor corresponding to a preset system throughput from a cloud desktop database; Based on the system response time and system throughput at each time node in each monitoring cycle, a comprehensive analysis is performed to obtain the cloud desktop system performance evaluation index.

4. The cloud desktop scheduling method according to claim 2, characterized in that: The comprehensive analysis obtains the predicted usage of various resources at each time node. The specific analysis process is as follows: Extract the preset cloud desktop user stickiness, cloud desktop system performance evaluation index and predicted usage evaluation impact factors corresponding to the actual usage of various resources from the cloud desktop database; Based on user stickiness, system performance evaluation index and the actual usage of various resources, a comprehensive analysis is conducted to obtain the predicted usage of various resources at each time node.

5. The cloud desktop scheduling method according to claim 1, characterized in that: The resource usage prediction accuracy index is a quantitative index obtained by comparing and analyzing the predicted usage and actual usage of each resource at each time node, and is used to quantify the accuracy of resource usage prediction.

6. The cloud desktop scheduling method according to claim 1, characterized in that: The cloud desktop system performance evaluation index is a quantitative indicator obtained by analyzing the system response time and system throughput of each time node in each monitoring cycle, and is used to quantify the performance of the cloud desktop system.

7. The cloud desktop scheduling method according to claim 1, characterized in that: The cloud desktop scheduling method is evaluated according to the resource usage prediction accuracy index, and the specific evaluation process is as follows: A resource usage prediction accuracy threshold is extracted from the cloud desktop database, and a resource usage prediction accuracy index is compared with the resource usage prediction accuracy threshold. If the resource usage prediction accuracy index is greater than or equal to the resource usage prediction accuracy threshold, the cloud desktop scheduling method is evaluated as qualified; if the resource usage prediction accuracy index is less than the resource usage prediction accuracy threshold, the cloud desktop scheduling method is evaluated as unqualified.

8. The cloud desktop scheduling method according to claim 2, characterized in that: The predicted usage of each resource at each time node is expressed as follows: Among them, y ik represents the predicted usage of the kth resource at the i-th time node, x ijk represents the usage of the kth resource at the i-th time node on the j-th day, σ ik represents the resource prediction compensation parameter of the kth resource at the ith time node, ω1 represents the prediction usage evaluation impact factor corresponding to the set historical average usage, ω2 represents the prediction usage evaluation impact factor corresponding to the set resource prediction compensation parameter, i represents the number of time nodes, i=1,2,3,...,n, n represents the total number of time nodes, j represents the number of monitoring cycles, j=1,2,3,...,m, m represents the total number of monitoring cycles, k represents the number of resources, k=1,2,3,...,h, h represents the total number of resources.

9. The cloud desktop scheduling method according to claim 2, characterized in that: The prediction accuracy index, the specific numerical expression is: Among them, C represents the prediction accuracy index, e represents the natural constant, and y ik represents the predicted usage of the kth resource at the i-th time node, represents the actual usage of the kth resource at the i-th time node, Δy represents the set allowable deviation usage, i represents the number of time nodes, i=1,2,3,...,n, n represents the total number of time nodes, k represents the number of resources, k=1,2,3,...,h, h represents the total number of resources.

10. A system using the cloud desktop scheduling method according to any one of claims 1 to 9, characterized in that: include: The cloud desktop data collection module is used to collect user behavior data, actual usage of various resources and system performance data. User behavior data includes the cumulative number of user logins, cumulative login duration and system response time; system performance data includes system throughput and resource usage; The cloud desktop predicted usage analysis module is used to process user behavior data to obtain the cloud desktop user stickiness, and process system performance data to obtain the cloud desktop system performance evaluation index. Combined with the actual usage of various resources, a comprehensive analysis is performed to obtain the predicted usage of various resources at each time node; The cloud desktop prediction accuracy analysis module is used to divide the time nodes according to the current time node, obtain each historical time node and each future time node, and obtain the resource usage prediction accuracy index based on the predicted usage of each type of resources and the actual usage of each type of resources at each historical time node through comprehensive analysis; The cloud desktop scheduling method evaluation module is used to evaluate the resource usage prediction results according to the resource usage prediction accuracy index. If the evaluation result is qualified, the cloud desktop resource scheduling is performed according to the various predicted usages at each future time node. If the evaluation result is unqualified, the resource usage prediction is performed again. The cloud desktop database is used to store cloud desktop data, which includes cloud desktop user stickiness evaluation influencing factors corresponding to preset cumulative login times and cumulative login duration, cloud desktop system performance evaluation influencing factors corresponding to preset system response time and preset system throughput, resource usage and predicted usage evaluation influencing factors corresponding to resource prediction compensation parameters, and resource usage prediction accuracy threshold.

Images