Computing power resource scheduling method and system in cloud environment

Abstract

The invention relates to the technical field of computing power resource scheduling, in particular to a computing power resource scheduling method and system in a cloud environment, and the method comprises the following steps: continuously collecting the utilization rates of a CPU and a GPU and network bandwidth data based on monitored cloud environment data, carrying out the grouping and statistics of resource use data, and carrying out the calculation of the resource use data; and determining an occupation proportion and a fluctuation condition of each computing resource. According to the method, the utilization condition and the demand change of each resource can be responded in real time by adopting continuous data monitoring and analysis, the resource competition state is calculated through dynamic analysis in resource allocation, the higher matching degree between tasks and resources is achieved, resource redundancy and waste are effectively reduced, overall scheduling is optimized, and the scheduling efficiency is improved. Through the strategy of monitoring the workload and adjusting the allocation state of the computing resources in real time, resource scheduling can adapt to the change of the computing power demand, the stability and efficiency of operation are enhanced, and the overall utilization rate of the computing resources is improved through deep analysis and optimal allocation of the use of the computing resources.

Description

Technical Field

[0001] The present invention relates to the technical field of computing power resource scheduling, and particularly to a computing power resource scheduling method and system in a cloud environment. Background Art

[0002] Computing power resource scheduling is an important technical field in computer science, which particularly focuses on how to effectively allocate and manage computing resources in a cloud computing environment. It involves various technologies, including load balancing, resource allocation algorithms, virtualization technologies, and the application of prediction and monitoring systems. The purpose of computing power resource scheduling is to optimize resource usage, reduce energy consumption, and improve the overall performance and reliability of the system, which is particularly important for large-scale data centers, multi-tenant cloud platforms, and high-performance computing environments.

[0003] Among them, the computing power resource scheduling method in a cloud environment refers to how to effectively allocate and manage computing resources to meet the needs of different users and applications under the cloud computing framework. The main purpose of such scheduling methods is to ensure that computing tasks can obtain the required resources, while optimizing resource utilization (load balancing) and cost-effectiveness, including dynamically allocating CPU, memory, storage, and network resources to support the concurrent needs of multi-tasks and multi-users.

[0004] When dealing with the highly variable cloud environment requirements in the prior art, the common static resource allocation strategies lack sufficient flexibility. In the traditional mode, computing resources are allocated according to fixed ratios or preset rules, which limits the ability to respond to emergencies. The lack of real-time data analysis results in the inability to accurately predict and respond to fluctuations in computing resource requirements, easily causing computing resource overload or idle, affecting system performance and user experience. The existing resource scheduling mechanisms do not support highly dynamic resource management, which is particularly disadvantageous in a concurrent environment of multi-users and multi-tasks, easily leading to a decline in service quality and affecting the operation efficiency and computing power control of the data center. Summary of the Invention

[0005] In order to solve the technical problems existing in the prior art that when dealing with the highly variable cloud environment requirements, the common static resource allocation strategies lack sufficient flexibility. In the traditional mode, computing resources are allocated according to fixed ratios or preset rules, which limits the ability to respond to emergencies. The lack of real-time data analysis results in the inability to accurately predict and respond to fluctuations in computing resource requirements, easily causing computing resource overload or idle, affecting system performance and user experience. The existing resource scheduling mechanisms do not support highly dynamic resource management, which is particularly disadvantageous in a concurrent environment of multi-users and multi-tasks, easily leading to a decline in service quality and affecting the operation efficiency and computing power control of the data center, the embodiments of the present invention provide a computing power resource scheduling method and system in a cloud environment. The technical solution is as follows: On the one hand, a computing power resource scheduling method in a cloud environment is provided, including the following steps: S1: Based on the monitored cloud environment data, continuously collect the usage rates of CPUs and GPUs and network bandwidth data, group and statistically analyze the resource usage data, determine the occupancy ratio and its fluctuation of each computing resource, and evaluate the competition status among resources to obtain the computing resource competition status; S2: Based on the computing resource competition status, group the computing tasks by priority, analyze the computing power and bandwidth requirements of the tasks, optimize the task priority and resource occupancy ratio, and at the same time compare the task requirements with the current resource status for real-time correction to obtain the computing resource allocation configuration; S3: Based on the computing resource allocation configuration, continuously monitor the workload of the current cloud platform, adjust the allocation status of computing resources according to the number of task queues and the fluctuation data of resource usage, match the real-time changes of the workload, and at the same time update the parameters of computing nodes and storage nodes to obtain the resource configuration update result; S4: Based on the resource configuration update result, compare the resource occupancy trend with the time interval of resource allocation adjustment, evaluate the load-demand ratio of computing nodes, perform resource allocation operations on computing nodes, optimize the resource distribution of the cloud environment, and obtain the resource allocation efficiency information; S5: Based on the resource allocation efficiency information, analyze the usage efficiency of each resource, adjust the priority and allocation ratio of computing resources by comparing the occupancy time with the average usage rate of the resources to obtain an overview of resource utilization scheduling.

[0006] On the other hand, the computing resource competition status includes resource utilization peaks, competition degree evaluation results, and key resource identification results. The computing resource allocation configuration specifically includes task-resource matching degree, priority adjustment index, and resource allocation efficiency. The resource configuration update result includes dynamic adjustment feedback information, node performance indicators, and resource reconfiguration rate. The resource allocation efficiency information specifically includes allocation response time, resource utilization balance degree, and efficiency improvement points. The resource utilization scheduling overview includes resource full-load rate, usage efficiency improvement result, and scheduling period optimization result.

[0007] On the other hand, the steps of continuously collecting the usage rates of CPUs and GPUs and network bandwidth data based on the monitored cloud environment data, grouping and statistically analyzing the resource usage data, determining the occupancy ratio and its fluctuation of each computing resource, and evaluating the competition status among resources to obtain the computing resource competition status are specifically as follows: S101: Based on the monitored cloud environment data, continuously collect the usage rates of CPUs and GPUs and network bandwidth data, divide the collected information by resource type, and analyze the usage frequency and fluctuation pattern of each type of resource to obtain an overview of resource usage; S102: Based on the resource usage overview, calculate the occupancy ratio of each computing resource in different time periods, determine the peak and trough periods of resource usage by analyzing the periodic fluctuations, and obtain the resource usage peak and trough diagram; S103: Based on the resource usage peak and trough diagram, analyze the usage conflicts between different computing resources, determine the peak usage data and idle period data of the resources, and identify the competition intensity of the computing resources, so as to obtain the computing resource competition status.

[0008] On the other hand, based on the computing resource competition status, group the computing tasks according to the priority, analyze the computing power and bandwidth requirements of the tasks, optimize the task priority and resource occupancy ratio, and at the same time compare the task requirements with the current resource status for real-time correction. The specific steps for obtaining the computing resource allocation configuration are as follows: S201: Based on the computing resource competition status, sort the computing tasks according to the priority, analyze the computing power and bandwidth requirements of each task, and adjust the task classification according to the resource competition data to obtain the task classification and demand comparison information; S202: Based on the task classification and demand comparison information, adjust the priority and resource ratio in real time, and repeatedly optimize the resource configuration according to the current resource status and task requirements to obtain the optimized resource allocation information; S203: Based on the optimized resource allocation information, compare the task requirements with the resource supply situation, verify item by item whether the resource supply meets the task requirements, analyze the deviation ratio in the allocation status, determine the current status of the task allocation, and obtain the computing resource allocation configuration.

[0009] On the other hand, when comparing the task requirements with the resource supply situation and verifying item by item whether the resource supply meets the task requirements, the formula is used: ; Analyze the deviation ratio in the allocation status, determine the current status of the task allocation, and obtain the computing resource allocation configuration, where represents the deviation ratio, represents the amount of computing power resources allocated to the th computing task in the cloud environment, represents the computing power demand of the th computing task, represents the total number of computing tasks.

[0010] On the other hand, based on the computing resource allocation configuration, monitor the current workload of the cloud platform in real time, adjust the allocation status of the computing resources according to the number of task queues and the fluctuation data of resource usage, match the real-time changes of the workload, and at the same time update the parameters of the computing nodes and storage nodes. The specific steps for obtaining the resource configuration update result are as follows: S301: Based on the computing resource allocation configuration, collect the workload data of the current cloud platform, screen the nodes critical for computing load through the number of task queues, statistically analyze the fluctuation range of resource usage, and analyze the bandwidth and computing power distribution required by tasks, and dynamically allocate the resource occupancy ratio to obtain the dynamic response information of the workload; S302: Based on the dynamic response information of the workload, update the configuration parameters of the computing nodes and storage nodes one by one, judge the resource supply-demand deviation of each node, and dynamically adjust the parameters of each node to match the current resource requirements to obtain the node resource adjustment result; S303: Based on the node resource adjustment result, verify the resource allocation status, adjust the resource occupancy difference and correct the allocation ratio by checking the matching degree between the resource allocation information and the real-time demand to obtain the resource configuration update result.

[0011] On the other hand, based on the resource configuration update result, compare the resource occupancy trend with the time interval of resource allocation adjustment, evaluate the load-demand ratio of the computing nodes, perform resource allocation operations on the computing nodes, and optimize the resource distribution of the cloud environment. The steps for obtaining the resource allocation efficiency information are specifically as follows: S401: Based on the resource configuration update result, collect the resource occupancy data, compare the historical occupancy trend with the current occupancy status, evaluate the resource allocation time interval, and determine the response speed of resource adjustment to obtain the resource adjustment response analysis result; S402: Based on the resource adjustment response analysis result, calculate the load-demand ratio of the computing nodes one by one, analyze whether the task resource occupancy reaches equilibrium, and match the resource requirements by adjusting the node allocation ratio. At the same time, optimize the task queue load and the resource distribution of the cloud environment to obtain the resource allocation efficiency information.

[0012] On the other hand, for calculating the load-demand ratio of the computing nodes one by one, analyzing whether the task resource occupancy reaches equilibrium, and matching the resource requirements by adjusting the node allocation ratio, optimizing the task queue load and the resource distribution of the cloud environment, the formula is used: ; To obtain the resource allocation efficiency information, where represents the resource occupancy balance of node , represents the load-demand ratio of node , which is calculated as , in the formula, is the current load of node , is the total demand of node , is the dynamic adjustment coefficient of node , Denote the average load demand ratio of all nodes, which represents the total number of resource nodes in the cloud environment.

[0013] On the other hand, based on the resource allocation efficiency information, analyze the usage efficiency of each resource. By comparing the occupation time with the average usage rate of the resource, adjust the priority and allocation ratio of computing resources. The specific steps to obtain the resource utilization scheduling overview are as follows: S501: Based on the resource allocation efficiency information, collect the occupation time and usage records of computing resources, compare the occupation time with the average usage rate of the corresponding resource item by item, identify the resource intervals with low efficiency, and obtain the resource usage efficiency overview; S502: Based on the resource usage efficiency overview, evaluate the priority and allocation ratio of computing resources, dynamically adjust the priority according to the resource requirements under different load states, reallocate the resource ratio, eliminate the supply-demand deviation, and obtain the resource ratio adjustment direction; S503: Based on the resource ratio adjustment direction, verify the resource allocation status node by node. By comparing the task load with the node resource allocation information, correct the allocation parameters to obtain the resource utilization scheduling overview.

[0014] On the other hand, a computing power resource scheduling system in a cloud environment is provided. This system is applied to the computing power resource scheduling method in a cloud environment and includes: The resource monitoring module continuously collects the usage rates of CPUs and GPUs and network bandwidth data based on the monitored cloud environment data, determines the occupation ratio and its fluctuation of each computing resource, and evaluates the competition status among resources to obtain the computing resource competition status; The task priority module groups computing tasks by priority based on the computing resource competition status, analyzes the computing power and bandwidth requirements of the tasks, and at the same time compares the task requirements with the current resource status for real-time correction to obtain the computing resource allocation configuration; The resource update module continuously monitors the workload of the current cloud platform based on the computing resource allocation configuration. According to the number of task queues and the fluctuation data of resource usage, adjust the allocation status of computing resources, and at the same time update the parameters of computing nodes and storage nodes to obtain the resource configuration update result; The resource allocation module compares the resource occupation trend with the time interval of resource allocation adjustment based on the resource configuration update result, evaluates the load-demand ratio of computing power nodes, and performs resource allocation operations on computing nodes to obtain the resource allocation efficiency information; The efficiency analysis module analyzes the usage efficiency of each resource based on the resource allocation efficiency information. By comparing the occupation time with the average usage rate of the resource, adjust the priority and allocation ratio of computing resources to obtain the resource utilization scheduling overview.

[0015] The beneficial effects brought by the technical solutions provided in the embodiments of the present invention at least include: By adopting continuous data monitoring and analysis, it is ensured that the utilization situation and demand changes of each resource can be immediately responded to. The resource allocation dynamically analyzes and calculates the resource competition state, achieving a higher matching degree between tasks and resources, effectively reducing resource redundancy and waste, and then optimizing the overall scheduling. The strategy of real-time monitoring of the workload and adjusting the calculation resource allocation status enables the resource scheduling to adapt to the changes in computing power requirements, enhancing the stability and efficiency of operation. Through in-depth analysis and optimized allocation of the use of computing resources, the overall utilization rate of computing resources is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0017] Figure 1 is the main step flow chart of the present invention; Figure 2 is the step flow chart of S1 of the present invention; Figure 3 is the step flow chart of S2 of the present invention; Figure 4 is the step flow chart of S3 of the present invention; Figure 5 is the step flow chart of S4 of the present invention; Figure 6 is the step flow chart of S5 of the present invention; Figure 7 is the system block diagram of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] The following will describe the technical solutions in the present invention with reference to the drawings.

[0019] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to represent examples, illustrations or explanations. Any embodiment or design solution described as "example" in the present invention should not be construed as being more preferred or having more advantages than other embodiments or design solutions. Exactly, the use of the word "example" is intended to present concepts in a specific way. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one of the two.

[0020] In the embodiments of the present invention, "image" and "picture" can sometimes be used interchangeably. It should be noted that when the difference is not emphasized, their intended meanings are the same. "Of", "corresponding", and "corresponding" can sometimes be used interchangeably. It should be noted that when the difference is not emphasized, their intended meanings are the same.

[0021] In the embodiments of the present invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, their intended meanings are the same.

[0022] To make the technical problems, technical solutions, and advantages to be solved by the present invention clearer, the following will be described in detail with reference to the accompanying drawings and specific embodiments.

[0023] The embodiments of the present invention provide a method for scheduling computing power resources in a cloud environment, as Figure 1 shown, including the following steps: S1: Based on the monitored cloud environment data, continuously collect the utilization rates of CPUs and GPUs and network bandwidth data, group and statistically analyze the resource usage data, determine the occupancy ratio and its fluctuation of each computing resource, and evaluate the competition state between resources to obtain the computing resource competition state; S2: Based on the computing resource competition state, group the computing tasks according to priority, analyze the computing power and bandwidth requirements of the tasks, optimize the task priority and resource occupancy ratio, and at the same time compare the task requirements with the current resource state for real-time correction to obtain the computing resource allocation configuration; S3: Based on the computing resource allocation configuration, continuously monitor the workload of the current cloud platform, adjust the allocation state of computing resources according to the number of task queues and the fluctuation data of resource usage, match the real-time changes of the workload, and at the same time update the parameters of computing nodes and storage nodes to obtain the resource configuration update result; S4: Based on the resource configuration update result, compare the resource occupancy trend with the time interval of resource allocation adjustment, evaluate the load-to-demand ratio of computing power nodes, perform resource allocation operations on computing nodes, and optimize the resource distribution of the cloud environment to obtain the resource allocation efficiency information; S5: Based on the resource allocation efficiency information, analyze the usage efficiency of each resource, and adjust the priority and allocation ratio of computing resources by comparing the occupancy time with the average utilization rate of resources to obtain an overview of resource utilization scheduling.

[0024] The computing resource competition status includes the resource utilization peak, the evaluation result of the competition degree, and the identification result of key resources. The specific computing resource allocation and configuration are the task-resource matching degree, the priority adjustment index, and the resource allocation efficiency. The resource configuration update result includes the dynamic adjustment feedback information, the node performance index, and the resource reallocation rate. The resource allocation efficiency information is specifically the allocation response time, the resource utilization balance degree, and the efficiency improvement point. The resource utilization scheduling overview includes the resource full load rate, the result of the usage efficiency improvement, and the result of the scheduling cycle optimization.

[0025] As Figure 2 shown, based on the monitored cloud environment data, continuously collect the usage rates of CPUs, GPUs, and network bandwidth data, group and statistically analyze the resource usage data, determine the occupancy ratio of each computing resource and its fluctuation situation, and evaluate the competition status among resources. The steps to obtain the computing resource competition status are specifically as follows: S101: Based on the monitored cloud environment data, continuously collect the usage rates of CPUs, GPUs, and network bandwidth data, divide the collected information by resource type, and analyze the usage frequency and fluctuation patterns of each type of resource to obtain an overview of resource usage; For the usage rates of CPUs, GPUs, and network bandwidth data, clarify the sampling interval and data storage format, perform real-time acquisition operations on the monitoring devices, and group and store the collected resource data by category through the set classification rules. The rules can include resource identifiers, timestamps, and related performance parameters. During the acquisition process, filter the data to eliminate invalid data or outliers. By statistically analyzing the time series data of each type of resource, calculate the average usage rate, peak usage rate, and fluctuation range of each resource. At the same time, analyze the resource usage frequency and change patterns, such as identifying the key patterns of data changes through the fluctuation range or periodic characteristics. Finally, generate a tabular or graphical overview of resource usage from the processed data.

[0026] S102: Based on the overview of resource usage, calculate the occupancy ratio of each computing resource in different time periods, and determine the peak and trough periods of resource usage by analyzing the periodic fluctuations to obtain a peak-trough diagram of resource usage; Divide the data into several intervals by time period, such as daytime, nighttime, etc. By summarizing and averaging the resource usage information within each interval, calculate the resource occupancy ratio within each interval. Use this data to further extract the fluctuation characteristics of resource usage, analyze the occupancy situation of each time period, and summarize the peak and trough periods of resource usage by comparing the usage data over multiple time cycles. After summarizing the results, generate a peak-trough diagram of resource usage to clarify the usage characteristics of resources in different time periods and provide a basis for subsequent scheduling operations.

[0027] S103: Based on the resource usage peak-valley diagram, analyze the differences to calculate the usage conflicts between resources, determine the peak usage data and idle period data of resources, and identify and calculate the competition intensity of resources to obtain the competition status of computing resources; Independently extract the usage data of various computing resources during peak and idle periods. By comparing the usage situations of different computing resources during peak periods, analyze the degree of usage conflicts between resources. According to the overlapping degree of resource usage duration and the change in occupancy ratio, define the competition situation of resources. By extracting the resource usage details during peak and trough periods, analyze the mutual influence of resources on usage requirements. Combine the peak usage data to identify the periods with strong resource competition, and combine the idle period data to clarify the optimizable resource allocation plan, and finally obtain the competition status of computing resources.

[0028] As Figure 3 shown, based on the competition status of computing resources, group the computing tasks according to priority, analyze the computing power and bandwidth requirements of the tasks, optimize the task priority and resource occupancy ratio, and at the same time compare the task requirements with the current resource status for real-time correction. The specific steps for obtaining the computing resource allocation configuration are as follows: S201: Based on the competition status of computing resources, sort the computing tasks by priority, analyze the computing power and bandwidth requirements of each task, and adjust the task classification according to the resource competition data to obtain the task classification and requirement comparison information; Classify the computing tasks according to resource requirements and priority parameters, analyze the computing power requirements and bandwidth requirements of each task, collect the performance records of each task during historical operations, group the computing power requirements according to task types, such as high-computation tasks, medium-computation tasks, and low-computation tasks. By retrieving the resource competition data, calculate the resource usage distribution of different task types during peak and trough periods. Combine the historical execution time and completion time ratio of the tasks to evaluate the importance of the tasks and assign priorities. Sort the tasks according to the priority level. At the same time, compare the computing power requirement data with the resource competition status. According to the resource occupancy conflict situation, adjust the task classification result and update the corresponding information of task types and resource requirements again, and finally form the task classification and requirement comparison information.

[0029] S202: Based on the task classification and requirement comparison information, adjust the priority and resource ratio in real time. According to the current resource status and task requirements, repeatedly optimize the resource configuration to obtain the optimized resource allocation information; Monitor the current resource status and task requirements in real time, retrieve the current task queue and resource usage data, group and sort tasks according to priority, and allocate preliminary resource ratios. Use a dynamic adjustment method to monitor the resource usage and remaining resource capacity during task execution in real time, compare the actual resource consumption and demand ratio of tasks, analyze the resource waste or deficiencies in the current allocation strategy, and adjust the allocation ratio of computing resources, such as increasing the resource amount for high-priority tasks and reducing the resource amount for low-priority tasks. Iteratively optimize the current resource configuration plan, record relevant performance parameters for each adjusted allocation plan, and generate optimized resource allocation information.

[0030] S203: Based on the optimized resource allocation information, compare the task requirements with the resource supply situation, verify item by item whether the resource supply meets the task requirements, analyze the deviation ratio in the allocation status, determine the current state of task allocation, and obtain the computing resource allocation configuration; Compare the task requirements and resource supply situation item by item, check whether the resources currently obtained by each task can meet the requirements, calculate the deviation ratio of the task requirements by comparing the computing power and bandwidth requirements of the tasks with the actual supply values, and classify and analyze the deviation ratio, such as dividing it into three categories: fully met, partially met, and not met. Record the current resource allocation status for each category of tasks, and verify and evaluate the resource allocation situation in combination with the deviation data of the optimization results to determine the current state of the computing resource allocation configuration.

[0031] Compare the task requirements with the resource supply situation, verify item by item whether the resource supply meets the task requirements, and use the formula: ; Analyze the deviation ratio in the allocation status, determine the current state of task allocation, and obtain the computing resource allocation configuration, where represents the deviation ratio, represents the computing power resource amount allocated to the th computing task in the cloud environment, represents the computing power demand of the th computing task, represents the total number of computing tasks; There are three computing tasks in the cloud environment. The computing power demand and the actual allocated computing power of each task are as follows. The data is obtained through the real-time monitoring system of the cloud management platform: , ; , ; , ; Calculate the deviation ratio for each task: Deviation ratio of the first task: ; Deviation ratio of the second task: ; Deviation ratio of the third task: ; Calculate the total deviation ratio: ; This result indicates that there is a deviation ratio of 0.444, i.e., 44.4%, in the scheduling of computing power resources in the cloud environment, manifested as over-allocation of resources for some tasks while some fail to meet the requirements, which helps to understand the imbalance in resource allocation, further optimize the resource allocation strategy, ensure more reasonable resource allocation, and improve the execution efficiency of computing tasks and resource utilization rate in the cloud environment.

[0032] As Figure 4 shown, based on the computing resource allocation configuration, the current workload of the cloud platform is monitored in real time, and according to the number of task queues and the fluctuation data of resource usage, the allocation status of computing resources is adjusted to match the real-time changes of the workload. At the same time, the parameters of computing nodes and storage nodes are updated, and the steps to obtain the updated result of resource configuration are specifically as follows: S301: Based on the computing resource allocation configuration, collect the workload data of the current cloud platform, screen the nodes critical for computing load through the number of task queues, count the fluctuation range of resource usage, analyze the bandwidth and computing power distribution required by tasks, and dynamically allocate the resource occupancy ratio to obtain the dynamic response information of the workload; Collect the current workload data of the cloud platform, including the real-time task queue length, the number of active tasks, and the resource usage of each node. By screening the tasks with a higher resource occupancy ratio in the task queue, locate the nodes critical for computing load, independently count the data such as the computing power, bandwidth, and memory usage of the nodes, obtain the fluctuation range and fluctuation frequency of resource usage of each node, analyze the distribution of bandwidth requirements and computing power requirements of the tasks on the nodes, dynamically adjust the resource allocation of the nodes according to the demand intensity, determine the adjustment of the allocation ratio by calculating the ratio of the remaining available resources to the actual demand, and record the allocation result after each dynamic adjustment. Finally, obtain the dynamic response information of the workload.

[0033] S302: Based on the dynamic response information of the workload, update the configuration parameters of the computing nodes and storage nodes one by one, judge the resource supply-demand deviation of each node, and dynamically adjust the parameters of each node to match the current resource demand to obtain the node resource adjustment result; Update the configuration parameters of the computing nodes and storage nodes, retrieve the resource supply-demand deviation information of each node one by one, determine the direction of parameter adjustment by analyzing the difference between the actual available resources of the node and the current task requirements. For example, reduce the resource allocation for non-critical tasks for high-load nodes, or allocate more high-priority tasks to idle nodes. At the same time, adjust the configuration parameters of the nodes in real time according to the deviation ratio. During the adjustment process, call the storage management module to update the allocation strategy of the storage nodes. The computing nodes achieve parameter update by adjusting the number of computing threads or restricting the task bandwidth occupancy ratio. Repeatedly adjust and monitor the change of resource occupancy until the supply-demand deviation is within a reasonable range, and finally record the adjustment results of the node resources.

[0034] S303: Based on the node resource adjustment results, verify the resource allocation status. By checking the matching degree between the resource allocation information and the real-time requirements, adjust the resource occupancy difference and correct the allocation ratio to obtain the resource configuration update results; Verify the resource allocation status item by item, compare the resource allocation information with the real-time requirements of the tasks, extract the real-time requirement data of each task and compare it with the allocated resource amount. According to the comparison results, judge the error range of resource allocation, analyze the over-allocation or under-allocation of resources in the allocation status, adjust the resource occupancy of each task one by one, correct the allocation ratio by reducing the resources of low-priority tasks or reallocating idle resources, record the resource status and task allocation situation of each node after correction, and generate the update results of the resource configuration.

[0035] As Figure 5 shown, based on the resource configuration update results, compare the resource occupancy trend with the time interval of resource allocation adjustment, evaluate the load-demand ratio of the computing nodes, and perform resource allocation operations on the computing nodes to optimize the resource distribution in the cloud environment. The steps to obtain the resource allocation efficiency information are specifically as follows: S401: Based on the resource configuration update results, collect resource occupancy data, compare the historical occupancy trend with the current occupancy status, evaluate the resource allocation time interval, and determine the response speed of resource adjustment to obtain the resource adjustment response analysis results; Collect the current resource occupancy data, including the resource utilization rate, task distribution, and occupancy duration of each computing node and storage node. At the same time, call the historical record module to extract the resource usage trend data in the approximate time period, compare the current status with the historical status, analyze the fluctuation amplitude and periodic changes of resource usage through time series data, compare the change speed of resource occupancy in each period, evaluate the reasonableness of the resource allocation interval, count the response time after the resource adjustment operation, and analyze the matching degree between the adjustment operation and the change of resource requirements in combination with the response time and occupancy change trend to obtain the resource adjustment response analysis results.

[0036] S402: Based on the analysis results of resource adjustment responses, calculate the load - to - demand ratio of each computing power node one by one, analyze whether the task resource occupancy reaches equilibrium, match the resource requirements by adjusting the node allocation ratio, and at the same time optimize the load of the task queue and the resource distribution in the cloud environment to obtain resource allocation efficiency information; Analyze the load - to - demand ratio of each computing power node one by one, extract the real - time load information of each node and match it with the task requirement list. By calculating the bandwidth occupancy ratio and computing resource consumption ratio in the task distribution of each node, judge whether the current load is balanced. For nodes with too high load, reduce low - priority tasks or transfer tasks to other idle nodes. For nodes with relatively low load, increase the allocated task volume or increase the resource allocation proportion of its high - priority tasks. At the same time, optimize the resource allocation of tasks waiting to be executed in the task queue, reduce the backlog phenomenon in the task queue by re - allocating resources, check and adjust the resource distribution of all nodes, and complete the optimal configuration of the overall resource distribution in the cloud environment.

[0037] Calculate the load - to - demand ratio of each computing power node one by one, analyze whether the task resource occupancy reaches equilibrium, and match the resource requirements by adjusting the node allocation ratio to optimize the load of the task queue and the resource distribution in the cloud environment. Use the formula: ; Obtain the resource allocation efficiency information, where represents the resource occupancy balance degree of node , represents the load - to - demand ratio of node , calculated as , where in the formula, is the current load of node , is the total demand of node , is the dynamic adjustment coefficient of node , represents the average load - to - demand ratio of all nodes, represents the total number of resource nodes in the cloud environment.

[0038] There are 3 nodes, and the situation is as follows: Node 1: Current load , total demand ; Node 2: Current load , total demand ; Node 3: Current load , total demand ; Based on the load data of the past 30 days, the standard deviations of the nodes are respectively: Node 1: 0.05, so ; Node 2: 0.10, so ; Node 3: 0.20, so ; Calculate the load demand ratio of each node and the average load demand ratio : ; ; ; ; Total number of nodes is 3.

[0039] Calculate the resource occupancy balance of each node : For Node 1: ; ; For Node 2: ; ; For Node 3: ; ; These results indicate that Node 3 has the highest resource occupancy balance due to its higher adjustment coefficient, indicating that its resource usage is relatively uneven. This calculation method helps to identify the nodes that need to be adjusted, thereby optimizing the resource allocation of the entire cloud environment.

[0040] As Figure 6 shown, based on the resource allocation efficiency information, analyze the usage efficiency of each resource, and by comparing the occupancy time with the average usage rate of the resource, the steps to adjust the priority and allocation ratio of computing resources to obtain an overview of resource utilization scheduling are specifically as follows: S501: Based on the resource allocation efficiency information, collect the occupancy time and usage records of computing resources, compare the occupancy time with the average usage rate of the corresponding resource item by item, identify the resource intervals with low efficiency, and obtain an overview of resource usage efficiency; The collected data is segmented and statistically analyzed according to the time dimension. The resource utilization rate and task allocation data within each time period are extracted. By comparing the resource occupation time in each time period with its corresponding average utilization rate, the utilization rate fluctuation of node resources in different time periods is analyzed. The intervals with a long occupation time but a low average utilization rate are marked, and the task types and allocation parameters in this interval are extracted. The proportion and distribution of low-efficiency tasks in the interval are statistically analyzed. Combined with the node's historical task records, the resource intervals that affect the overall efficiency are further screened to generate an overview of resource utilization efficiency.

[0041] S502: Based on the overview of resource utilization efficiency, evaluate the priority and allocation ratio of computing resources, dynamically adjust the priority according to the resource requirements under different load states, reallocate the resource ratio, eliminate the supply-demand deviation, and obtain the direction of resource ratio adjustment; Analyze the priority and allocation ratio of each type of computing resource, extract the task allocation parameters of high-priority resources and low-priority resources, combine the resource requirements under different load states, dynamically adjust the resource priority. By retrieving the high-load task data in the task queue, increase the allocation ratio of high-priority resources for high-load tasks, and at the same time reduce the resources occupied by low-priority tasks. Compare the real-time resource requirements of high-load tasks with the current allocation situation of the node, calculate the resource ratio that needs to be adjusted, reduce the supply-demand difference through optimizing the reallocation of resources, update the resource allocation record of the node, and determine the direction of resource ratio adjustment.

[0042] S503: Based on the direction of resource ratio adjustment, verify the resource allocation status node by node. By comparing the task load and node resource allocation information, correct the allocation parameters to obtain an overview of resource utilization scheduling; Verify the resource allocation status of each node one by one. By extracting the task load and resource allocation record of each node, check the matching degree between the task load and the allocated resources. Correct the parameters for nodes with too high or too low load. By reducing the resource occupation of low-priority tasks or adjusting tasks to other idle nodes, correct the resource allocation ratio of each node. Combine the node allocation record and the task requirement list to complete the update of the allocation parameters, gradually balance the difference between resource allocation and actual load, and generate an overview of resource utilization scheduling.

[0043] As Figure 7 shown, a computing resource scheduling system in a cloud environment includes: The resource monitoring module continuously collects the utilization rates of CPUs and GPUs and network bandwidth data based on the monitored cloud environment data, determines the occupation ratio and its fluctuation of each type of computing resource, and evaluates the competition state between resources to obtain the computing resource competition state; The task priority module groups computing tasks by priority based on the computing resource competition status, analyzes the computing power and bandwidth requirements of the tasks, and at the same time compares the task requirements with the current resource status for real-time correction to obtain the computing resource allocation configuration; The resource update module monitors the current workload of the cloud platform in real time based on the computing resource allocation configuration, adjusts the allocation status of computing resources according to the number of task queues and the fluctuation data of resource usage, and at the same time updates the parameters of computing nodes and storage nodes to obtain the resource configuration update result; The resource allocation module compares the resource occupancy trend with the time interval of resource allocation adjustment based on the resource configuration update result, evaluates the load-demand ratio of computing power nodes, and performs resource allocation operations on computing nodes to obtain the resource allocation efficiency information; The efficiency analysis module analyzes the usage efficiency of each resource based on the resource allocation efficiency information, adjusts the priority and allocation ratio of computing resources by comparing the occupancy time with the average usage rate of resources, and obtains an overview of resource utilization scheduling.

[0044] It should be understood that the term "and / or" in this article is merely a description of the association relationship between associated objects, indicating that three relationships can exist. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. Here, A and B can be singular or plural. In addition, the character " / " in this article generally represents an "or" relationship between the associated objects before and after, but it may also represent an "and / or" relationship, which can be specifically understood by referring to the context.

[0045] In the present invention, "at least one" means one or more, and "a plurality" means two or more. "At least one (item)" or its similar expression means any combination of these items, including any combination of single item (item) or plural items (items). For example, at least one (item) of a, b, or c can represent: a, b, c, a - b, a - c, b - c, or a - b - c, where a, b, and c can be single or multiple.

[0046] It should be understood that in various embodiments of the present invention, the magnitudes of the sequence numbers of the above processes do not mean the sequence of execution. The execution sequence of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.

[0047] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0048] Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the devices, apparatuses, and units described above can refer to the corresponding processes in the foregoing method embodiments, and will not be elaborated herein.

[0049] In several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there can be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling, direct coupling, or communication connection to each other can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be in an electrical, mechanical, or other form.

[0050] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place, or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0051] In addition, the functional units in various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit.

[0052] When the above-mentioned functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM), random access memories (RAM), magnetic disks, or optical discs.

[0053] As described above, the above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. A computing power resource scheduling method in a cloud environment, characterized in that The method includes: S1: Based on the monitored cloud environment data, continuously collect the utilization rates of CPUs, GPUs, and network bandwidth data, group and statistically analyze the resource usage data, determine the occupancy ratio and its fluctuations of each computing resource, and evaluate the competition status among resources to obtain the computing resource competition status; S2: Based on the computing resource competition status, group the computing tasks by priority, analyze the computing power and bandwidth requirements of the tasks, optimize the task priorities and resource occupancy ratios, and at the same time compare the task requirements with the current resource status for real-time correction to obtain the computing resource allocation configuration; S3: Based on the computing resource allocation configuration, continuously monitor the workload of the current cloud platform, adjust the allocation status of computing resources according to the number of task queues and the fluctuation data of resource usage, match the real-time changes of the workload, and at the same time update the parameters of computing nodes and storage nodes to obtain the resource configuration update result; S4: Based on the resource configuration update result, compare the resource occupancy trend with the time interval of resource allocation adjustment, evaluate the load-to-demand ratio of computing power nodes, perform resource allocation operations on computing nodes, optimize the resource distribution of the cloud environment, and obtain the resource allocation efficiency information; S5: Based on the resource allocation efficiency information, analyze the usage efficiency of each resource, adjust the priorities and allocation ratios of computing resources by comparing the occupancy time with the average utilization rate of resources, and obtain the resource utilization scheduling overview.

2. The computing power resource scheduling method in the cloud environment according to claim 1, wherein The computing resource competition status includes resource utilization peaks, competition degree evaluation results, and key resource identification results. The computing resource allocation configuration specifically includes task-resource matching degrees, priority adjustment indices, and resource allocation efficiencies. The resource configuration update result includes dynamic adjustment feedback information, node performance indicators, and resource reconfiguration rates. The resource allocation efficiency information specifically includes allocation response times, resource utilization balance degrees, and efficiency improvement points. The resource utilization scheduling overview includes resource full-load rates, usage efficiency improvement results, and scheduling cycle optimization results.

3. The computing power resource scheduling method in the cloud environment according to claim 1, wherein, The steps for obtaining the computing resource competition status by continuously collecting the utilization rates of CPUs, GPUs, and network bandwidth data based on the monitored cloud environment data, grouping and statistically analyzing the resource usage data, determining the occupancy ratio and its fluctuations of each computing resource, and evaluating the competition status among resources are specifically as follows: S101: Based on the monitored cloud environment data, continuously collect the utilization rates of CPUs, GPUs, and network bandwidth data, divide the collected information by resource type, and analyze the usage frequencies and fluctuation patterns of each type of resource to obtain an overview of resource usage; S102: Based on the overview of resource usage, calculate the occupancy ratio of each computing resource in different time periods, and determine the peak and trough periods of resource usage by analyzing periodic fluctuations to obtain a resource usage peak-valley diagram; S103: Based on the resource usage peak-valley diagram, analyze the usage conflicts between different computing resources, determine the resource peak usage data and idle period data, and identify the competition intensity of computing resources to obtain the computing resource competition status.

4. The computing power resource scheduling method in the cloud environment according to claim 1, characterized in that Based on the computing resource competition status, group computing tasks by priority, analyze the computing power and bandwidth requirements of tasks, optimize task priorities and resource occupancy ratios, and at the same time compare task requirements with the current resource status for real-time correction. The specific steps for obtaining the computing resource allocation configuration are as follows: S201: Based on the computing resource competition status, sort the computing tasks by priority, analyze the computing power and bandwidth requirements of each task, and adjust the task classification according to the resource competition data to obtain the task classification and requirement comparison information; S202: Based on the task classification and requirement comparison information, adjust the priority and resource ratio in real time, and repeatedly optimize the resource configuration according to the current resource status and task requirements to obtain the optimized resource allocation information; S203: Based on the optimized resource allocation information, compare task requirements with resource supply conditions, verify item by item whether the resource supply meets the task requirements, analyze the deviation ratio in the allocation status, determine the current status of task allocation, and obtain the computing resource allocation configuration.

5. The computing power resource scheduling method in the cloud environment according to claim 4, characterized in that, When comparing task requirements with resource supply conditions and verifying item by item whether the resource supply meets the task requirements, the formula is used: ； Analyze the deviation ratio in the allocation status, determine the current status of task allocation, and obtain the computing resource allocation configuration, where represents the deviation ratio, represents the amount of computing power resources allocated to the th computing task in the cloud environment, represents the computing power demand of the th computing task, and represents the total number of computing tasks.

6. The computing power resource scheduling method in the cloud environment according to claim 1, wherein Based on the computing resource allocation configuration, monitor the workload of the current cloud platform in real time, adjust the allocation status of computing resources according to the number of task queues and the fluctuation data of resource usage, match the real-time changes of the workload, and at the same time update the parameters of computing nodes and storage nodes. The specific steps for obtaining the resource configuration update result are as follows: S301: Based on the computing resource allocation configuration, collect the workload data of the current cloud platform, screen the nodes critical for computing load through the number of task queues, count the fluctuation range of resource usage, and analyze the bandwidth and computing power distribution required by tasks to dynamically allocate the resource occupancy ratio to obtain the workload dynamic response information; S302: Based on the workload dynamic response information, update the configuration parameters of computing nodes and storage nodes one by one, judge the resource supply and demand deviation of each node, and dynamically adjust the parameters of each node to match the current resource requirements to obtain the node resource adjustment result; S303: Based on the node resource adjustment result, verify the resource allocation status, adjust the resource occupancy difference and correct the allocation ratio by checking the matching degree between the resource allocation information and the real-time requirements to obtain the resource configuration update result.

7. The computing power resource scheduling method in the cloud environment according to claim 1, characterized in that Based on the resource configuration update result, compare the resource occupancy trend with the time interval of resource allocation adjustment, evaluate the load-to-demand ratio of computing nodes, perform resource allocation operations on computing nodes, and optimize the resource distribution of the cloud environment. The specific steps for obtaining the resource allocation efficiency information are as follows: S401: Based on the resource configuration update result, collect resource occupancy data, compare the historical occupancy trend with the current occupancy status, evaluate the resource allocation time interval, and determine the response speed of resource adjustment to obtain the resource adjustment response analysis result; S402: Based on the resource adjustment response analysis results, calculate the load-to-demand ratio of computing nodes one by one, analyze whether the task resource occupancy reaches equilibrium, match the resource requirements by adjusting the node allocation ratio, and simultaneously optimize the task queue load and the resource distribution in the cloud environment to obtain resource allocation efficiency information.

8. The computing power resource scheduling method in the cloud environment according to claim 7, characterized in that, The steps of calculating the load-to-demand ratio of computing nodes one by one, analyzing whether the task resource occupancy reaches equilibrium, matching the resource requirements by adjusting the node allocation ratio, optimizing the task queue load and the resource distribution in the cloud environment adopt the formula: ； Obtain the resource allocation efficiency information, where represents the resource occupancy balance of node ; represents the load demand ratio of node , calculated as , where is the current load of node ; is the total demand of node ; is the dynamic adjustment coefficient of node ; represents the average load demand ratio of all nodes, and represents the total number of resource nodes in the cloud environment.

9. The computing power resource scheduling method in the cloud environment according to claim 1, wherein The steps of analyzing the usage efficiency of each resource based on the resource allocation efficiency information, adjusting the priority and allocation ratio of computing resources by comparing the occupancy time with the average usage rate of the resources to obtain an overview of resource utilization scheduling are specifically as follows: S501: Based on the resource allocation efficiency information, collect the occupancy time and usage records of computing resources, compare the occupancy time with the average usage rate of the corresponding resources item by item, identify resource intervals with low efficiency, and obtain an overview of resource usage efficiency; S502: Based on the overview of resource usage efficiency, evaluate the priority and allocation ratio of computing resources, dynamically adjust the priority according to the resource requirements under different load states, reallocate the resource ratio, and eliminate the supply-demand deviation to obtain the resource ratio adjustment direction; S503: Based on the resource ratio adjustment direction, verify the resource allocation status for each node one by one, correct the allocation parameters by comparing the task load and the node resource allocation information, and obtain an overview of resource utilization scheduling.

10. A computing power resource scheduling system in a cloud environment, the computing power resource scheduling system in the cloud environment is used to implement the computing power resource scheduling method in the cloud environment according to any one of claims 1-9, characterized in that, The system includes: The resource monitoring module continuously collects the usage rates of CPUs, GPUs and network bandwidth data based on the monitored cloud environment data, determines the occupancy ratio and its fluctuation of each computing resource, and evaluates the competition status among resources to obtain the computing resource competition status; The task priority module groups computing tasks by priority based on the computing resource competition status, analyzes the computing power and bandwidth requirements of the tasks, and simultaneously compares the task requirements with the current resource status for real-time correction to obtain the computing resource allocation configuration; The resource update module continuously monitors the workload of the current cloud platform based on the computing resource allocation configuration, adjusts the allocation status of computing resources according to the number of task queues and the fluctuation data of resource usage, and simultaneously updates the parameters of computing nodes and storage nodes to obtain the resource configuration update result; The resource allocation module compares the resource occupancy trend with the time interval of resource allocation adjustment based on the resource configuration update result, evaluates the load-to-demand ratio of computing nodes, and performs resource allocation operations on computing nodes to obtain resource allocation efficiency information; The efficiency analysis module analyzes the usage efficiency of each resource based on the resource allocation efficiency information, adjusts the priority and allocation ratio of computing resources by comparing the occupancy time with the average usage rate of the resources to obtain an overview of resource utilization scheduling.

Images