A method and system for allocating and scheduling enterprise computing resources

Abstract

The present invention discloses a method and system for allocating and scheduling enterprise computing power resources, and specifically relates to the technical field of resource allocation and scheduling. Test tasks are generated by simulating real scenarios, and scheduling success rate fluctuation characteristics and conflict resolution efficiency characteristics are extracted. The scheduling accuracy is evaluated by a machine learning model, and the results are classified into accuracy scheduling, incomplete accuracy scheduling, and inaccuracy scheduling. For accuracy scheduling, the existing strategy is maintained and continuously monitored. For inaccuracy scheduling, the algorithm is reconstructed. For incomplete accuracy scheduling, the resource pool ratio is dynamically adjusted by predicting the degree of abnormality, and low-priority tasks are switched to a spare resource pool when the degree of abnormality is high, which significantly improves the rationality of resource allocation and scheduling accuracy, ensures the real-time performance and resource utilization efficiency of key tasks, and reduces the system scheduling failure rate and the economic losses caused by resource conflicts.

Description

Technical Field

[0001] The present invention relates to the technical field of resource allocation and scheduling, and in particular to a method and system for allocating and scheduling enterprise computing power resources. Background Art

[0002] Enterprise computing resource allocation and scheduling refers to the reasonable allocation, management, and dynamic adjustment of computing resources (such as CPU, GPU, memory, storage, and network bandwidth) within the enterprise or external cloud computing environment to meet the needs of different business scenarios, improve resource utilization, and ensure efficient completion of tasks.

[0003] In different business scenarios, the computing tasks of enterprises usually have different resource requirements. For example, data analysis tasks may require a CPU with high computing power, while machine learning training relies on GPU acceleration. The goal of resource allocation is to reasonably allocate limited computing resources to corresponding tasks or business units based on business needs and the priority of computing tasks to avoid resource waste or bottlenecks. For example, when multiple departments apply for resources at the same time, computing power can be allocated based on scheduling strategies to meet the real-time or performance requirements of key businesses. Scheduling is the core of the allocation process, which involves not only initial resource allocation, but also dynamic adjustments during task execution. Modern scheduling systems usually use AI and big data technologies to analyze current resource utilization, task queue status and business indicators, and optimize resource scheduling strategies in real time. For example, when resources are released after certain tasks are completed, the system can dynamically allocate these resources to other tasks, or temporarily call cloud computing expansion when resources are tight, thereby ensuring the efficient execution of the overall task.

[0004] The prior art has the following deficiencies:

[0005] When the system tries to allocate the released resources to other tasks, if there are multiple high-priority tasks competing for the same batch of resources at the same time, scheduling failure or unreasonable resource allocation may occur. Some critical tasks (such as real-time financial transaction processing or medical diagnosis systems) may be delayed or interrupted due to insufficient resources, resulting in serious economic losses. In addition, inaccurate scheduling can lead to a mismatch between resource allocation and task requirements, such as allocating too many resources to low-priority tasks, while high-priority tasks have insufficient resources. Idle resources cannot be used reasonably, and some tasks need to repeatedly apply for or wait for resources. High-priority tasks are stalled due to insufficient resources, and low-priority tasks may occupy critical computing power due to over-allocation. Summary of the invention

[0006] The purpose of the present invention is to provide a method and system for allocating and scheduling enterprise computing resources to address the deficiencies in the background technology.

[0007] To achieve the above object, the present invention provides the following technical solution: An enterprise computing power resource allocation and scheduling method, comprising the following steps:

[0008] S1: Create multiple test tasks, including tasks with different priorities and different resource requirements, simulate high-concurrency scenarios in a real environment, including several high-priority tasks competing for limited resources, define the resource pool capacity constraint, and at the same time set the scheduling algorithm used in the test scenario;

[0009] S2: Run the test according to the designed task queue and resource pool, monitor the resource allocation and task execution of the resource allocation and scheduling system, record the whole process of each task from resource request to allocation and execution completion, and extract the scheduling success rate fluctuation characteristics during the test process and the resolution efficiency characteristics when conflicts occur;

[0010] S3: According to the extracted scheduling success rate fluctuation characteristics and the resolution efficiency characteristics when conflicts occur, evaluate the accuracy of real-time resource scheduling for high-priority tasks in a high-concurrency environment;

[0011] S4: According to the evaluation results, divide the accuracy of real-time resource scheduling for high-priority tasks in a high-concurrency environment into accurate scheduling, incomplete accurate scheduling, and inaccurate scheduling;

[0012] S5: For accurate scheduling, maintain the current scheduling algorithm and strategy, and continuously monitor the resource allocation performance of the system; for inaccurate scheduling, perform algorithm replacement and reconstruction;

[0013] S6: For incomplete accurate scheduling, predict the degree of abnormality of the accuracy of real-time resource scheduling for high-priority tasks in a high-concurrency environment, dynamically adjust the resource pool allocation ratio according to the prediction results, and when the degree of abnormality is high, switch low-priority tasks to the standby resource pool to release resources for high-priority tasks to use.

[0014] Preferably, in S2, after analyzing the extracted scheduling success rate fluctuation characteristics, a scheduling success rate fluctuation index is generated, and the obtaining method of the scheduling success rate fluctuation index is:

[0015] Let the scheduling success rate be recorded by time period during the test, and the time series is: ; where represents the scheduling success rate of the nth time period, and n is the total number of time periods; calculate the moving average value of the scheduling success rate, and the formula is: ; is the moving average value of the i-th time period, w is the window size, is the scheduling success rate data point within the window. When i < w, only the first i data points are used to calculate the mean; calculate the scheduling success rate for the moving average value The degree of deviation , the formula is: ; In the formula, is the deviation of the scheduling success rate in the i-th time period, reflecting the difference between the current value and the moving average trend. The overall scheduling success rate fluctuation index reflects the fluctuation range of the entire time series. The formula is: ; Where W is the scheduling success rate fluctuation index.

[0016] Preferably, in S2, the conflict resolution efficiency anomaly index is generated after analyzing the extracted conflict resolution efficiency features. The conflict resolution efficiency anomaly index is obtained by:

[0017] Assume that the time series of conflict resolution efficiency is: ; is the efficiency of resolving the mth conflict, m is the total number of data points, calculate the 25% quantile Q1 and 75% quantile Q3 of the conflict resolution efficiency, as well as the interquartile range IQR, the formula is: Q1=⌊0.25×n⌋th data point; Q3=⌊0.75×n⌋th data point; IQR=Q3−Q1; set the upper and lower bounds of normal data according to the quantile range: lower bound=Q1−1.5×IQR; upper bound=Q3+1.5×IQR; mark the data points beyond the upper and lower bounds as abnormal points: if <Nether or > upper bound, then is an outlier point, and the ratio of the number of outliers to the total data points m is calculated, that is, the conflict resolution efficiency anomaly index is calculated.

[0018] Preferably, in S3, the accuracy of real-time resource scheduling of high-priority tasks in a high-concurrency environment is evaluated based on the extracted scheduling success rate fluctuation characteristics and the resolution efficiency characteristics when conflicts occur, specifically:

[0019] The scheduling success rate fluctuation index and the conflict resolution efficiency anomaly index are converted into comprehensive feature vectors, and the comprehensive feature vectors are used as the input of the machine learning model. The machine learning model uses each group of comprehensive feature vectors to predict the accuracy value label of real-time resource scheduling for high-priority tasks in a high-concurrency environment as the prediction target, and takes minimizing the sum of prediction errors of accuracy value labels for real-time resource scheduling of high-priority tasks in all high-concurrency environments as the training target. The machine learning model is trained until the sum of prediction errors converges, and the model training is stopped. The accuracy value of real-time resource scheduling for high-priority tasks in a high-concurrency environment is determined according to the model output results, wherein the machine learning model is a polynomial regression model.

[0020] Preferably, in S4, according to the evaluation result, the accuracy of real-time resource scheduling of high-priority tasks in a high-concurrency environment is divided into accuracy scheduling, incomplete accuracy scheduling and inaccuracy scheduling;

[0021] Compare the acquired accuracy value of real-time resource scheduling of high-priority tasks in a high-concurrency environment with a gradient standard threshold, where the gradient standard threshold includes a first standard threshold and a second standard threshold, and the first standard threshold is less than the second standard threshold, and compare the accuracy value of real-time resource scheduling of high-priority tasks in a high-concurrency environment with the first standard threshold and the second standard threshold respectively;

[0022] If the accuracy value of real-time resource scheduling of high-priority tasks in a high-concurrency environment is greater than the second standard threshold, it means that the accuracy of real-time resource scheduling of high-priority tasks in a high-concurrency environment is high, and a high-accuracy scheduling signal is generated at this time, and it is classified as accuracy scheduling;

[0023] If the accuracy value of real-time resource scheduling of high-priority tasks in a high-concurrency environment is greater than or equal to the first standard threshold and less than or equal to the second standard threshold, it means that the accuracy of real-time resource scheduling of high-priority tasks in a high-concurrency environment is medium, and a medium-accuracy scheduling signal is generated at this time, and it is divided into incomplete accuracy scheduling;

[0024] If the accuracy value of real-time resource scheduling of high-priority tasks in a high-concurrency environment is less than the first standard threshold, it means that the accuracy of real-time resource scheduling of high-priority tasks in a high-concurrency environment is low. At this time, a low-accuracy scheduling signal is generated and classified as inaccurate scheduling.

[0025] Preferably, in S6, for incomplete accuracy scheduling, the abnormal degree of accuracy of real-time resource scheduling of high-priority tasks in a high concurrency environment is predicted, and the resource pool allocation ratio is dynamically adjusted according to the prediction result, specifically: for incomplete accuracy scheduling, that is, the accuracy value of real-time resource scheduling of high-priority tasks in a high concurrency environment generated within a fixed time period is greater than or equal to the first standard threshold and less than or equal to the second standard threshold, the accuracy values ​​generated within a subsequent fixed time period that are greater than or equal to the first standard threshold and less than or equal to the second standard threshold are collected, and the corresponding time series is constructed, the mean and standard deviation of the time series are calculated, and after analysis, the abnormal degree of accuracy of real-time resource scheduling of high-priority tasks in a high concurrency environment is predicted.

[0026] Preferably, if the mean accuracy value in the data set is greater than or equal to the reference threshold of the mean accuracy value, and the standard deviation of the accuracy value is less than the reference threshold of the standard deviation of the accuracy value, the current resource pool allocation ratio should be maintained without large-scale adjustment, and the accuracy value should continue to be monitored to prevent fluctuations from increasing;

[0027] If the mean accuracy value is greater than or equal to the reference threshold of the mean accuracy value, and the standard deviation of the accuracy value is greater than or equal to the reference threshold of the standard deviation of the accuracy value, then the resource pool allocation ratio should be dynamically adjusted: more resources should be reserved for high-priority tasks, and the resource allocation priority of low-priority tasks should be reduced;

[0028] If the mean accuracy value is less than the reference threshold of the mean accuracy value, and the standard deviation of the accuracy value is greater than or equal to the reference threshold of the standard deviation of the accuracy value, the low-priority task should be switched to the backup resource pool, the main resource pool should be released to give priority to the high-priority task, and the dynamic priority adjustment algorithm should be used to timely increase the priority of the deadline task;

[0029] If the mean accuracy value is less than the reference threshold of the mean accuracy value, and the standard deviation of the accuracy value is less than the reference threshold of the standard deviation of the accuracy value, the resource allocation ratio of high-priority tasks should be increased, and the occupancy of medium and low-priority tasks should be reduced.

[0030] The present invention also provides an enterprise computing resource allocation and scheduling system, including a scenario configuration module, a feature extraction module, a scheduling accuracy evaluation module, a scheduling classification module, a scheduling strategy execution module and a dynamic adjustment module:

[0031] Scenario configuration module: creates multiple test tasks, including tasks of different priorities and different resource requirements, simulating high-concurrency scenarios in real environments, including several high-priority tasks competing for limited resources, defining resource pool capacity constraints, and setting the scheduling algorithm used in the test scenario;

[0032] Feature extraction module: Run the test according to the designed task queue and resource pool, monitor the resource allocation and task execution of the resource allocation and scheduling system, record the entire process of each task from resource request to allocation and execution completion, and extract the scheduling success rate fluctuation characteristics during the test process and the resolution efficiency characteristics when conflicts occur;

[0033] Scheduling accuracy evaluation module: evaluates the accuracy of real-time resource scheduling of high-priority tasks in a high-concurrency environment based on the extracted scheduling success rate fluctuation characteristics and conflict resolution efficiency characteristics;

[0034] Scheduling classification module: Based on the evaluation results, the accuracy of real-time resource scheduling for high-priority tasks in a high-concurrency environment is divided into accurate scheduling, incompletely accurate scheduling, and inaccurate scheduling;

[0035] Scheduling strategy execution module: For accurate scheduling, the current scheduling algorithm and strategy should be maintained, and the resource allocation performance of the system should be continuously monitored; for inaccurate scheduling, the algorithm should be replaced and reconstructed;

[0036] Dynamic adjustment module: For incomplete accuracy scheduling, the abnormal degree of real-time resource scheduling accuracy of high-priority tasks in a high-concurrency environment is predicted, and the resource pool allocation ratio is dynamically adjusted according to the prediction results. When the abnormal degree is high, low-priority tasks are switched to the backup resource pool to release resources for high-priority tasks.

[0037] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0038] 1. The present invention constructs a high-concurrency scenario, comprehensively considers the fluctuation characteristics of scheduling success rate and the efficiency characteristics of conflict resolution, uses a machine learning model to accurately evaluate the accuracy of real-time resource scheduling of high-priority tasks, and dynamically adjusts the resource pool allocation strategy. For different scheduling categories, the present invention adopts differentiated optimization schemes, such as maintaining existing strategies, dynamically allocating resources, or reconstructing scheduling algorithms, thereby solving the problems of scheduling failure, unreasonable resource allocation, insufficient resources for high-priority tasks, and low resource utilization in the prior art, and significantly improves the system's scheduling accuracy and reliability for high-concurrency tasks.

[0039] 2. The present invention can efficiently guarantee the real-time requirements of high-priority tasks in a high-concurrency environment, optimize resource allocation efficiency, and reduce economic losses caused by improper scheduling. At the same time, under the condition of incomplete accuracy scheduling, the present invention dynamically adjusts the resource pool allocation ratio by predicting the degree of abnormality, releases the main resource pool for key tasks, realizes intelligent and dynamic management of resource scheduling, ensures the stability and continuous optimization capability of the system, and meets the application needs of enterprises in complex computing power demand scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0041] Figure 1 The present invention is a flow chart of the method.

[0042] Figure 2 It is a system module diagram of the present invention. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0044] Example 1, please refer to Figure 1 and Figure 2 As shown, the enterprise computing resource allocation and scheduling method described in this embodiment includes the following steps:

[0045] S1: Create multiple test tasks, including tasks of different priorities and resource requirements, to simulate high-concurrency scenarios in real environments, including several high-priority tasks competing for limited resources, define resource pool capacity constraints, and set the scheduling algorithm used in the test scenario;

[0046] S2: Run the test according to the designed task queue and resource pool, monitor the resource allocation and task execution of the resource allocation and scheduling system, record the entire process of each task from resource request to allocation and execution completion, and extract the scheduling success rate fluctuation characteristics during the test process and the resolution efficiency characteristics when conflicts occur;

[0047] S3: Based on the extracted scheduling success rate fluctuation characteristics and conflict resolution efficiency characteristics, the accuracy of real-time resource scheduling of high-priority tasks in a high-concurrency environment is evaluated;

[0048] S4: Based on the evaluation results, the accuracy of real-time resource scheduling for high-priority tasks in a high-concurrency environment is divided into accurate scheduling, incompletely accurate scheduling, and inaccurate scheduling;

[0049] S5: For accurate scheduling, the current scheduling algorithm and strategy should be maintained, and the resource allocation performance of the system should be continuously monitored; for inaccurate scheduling, the algorithm should be replaced and reconstructed;

[0050] S6: For incomplete accuracy scheduling, predict the degree of abnormality in the accuracy of real-time resource scheduling of high-priority tasks in a high-concurrency environment, and dynamically adjust the resource pool allocation ratio based on the prediction results. When the degree of abnormality is high, switch low-priority tasks to the backup resource pool to release resources for high-priority tasks.

[0051] In S1, multiple test tasks are created, including tasks of different priorities and different resource requirements, to simulate high-concurrency scenarios in real environments, including several high-priority tasks competing for limited resources, defining resource pool capacity constraints, and setting the scheduling algorithm used in the test scenario, specifically:

[0052] The task set is designed to simulate high-concurrency scenarios in real environments, including different priorities and resource requirements, to ensure that the test covers critical situations where high-priority tasks compete for limited resources.

[0053] The properties set for each task should include: High Priority (HP): requires strong real-time guarantees, such as trading systems or real-time diagnosis. Medium Priority (MP): requires moderate real-time and resource guarantees, such as data analysis or video processing. Low Priority (LP): non-urgent tasks, such as log processing or offline training. Resource requirements: Number of CPU cores: the computing resources required for the task (such as 1-8 cores). Memory (RAM): the memory size required for the task to run (such as 2-32 GB). I / O requirements: the bandwidth requirements of the task for the disk or network (such as high, medium, low). Task duration: the estimated time for the task to run, combining short-term tasks (5-60 seconds) with long-term tasks (1-10 minutes).

[0054] Task distribution design includes: Task quantity: To simulate high concurrency scenarios, a total of 1,000 tasks are designed. Priority distribution: High priority tasks (HP): 30% (300). Medium priority tasks (MP): 40% (400). Low priority tasks (LP): 30% (300). Resource requirement distribution: High priority tasks: CPU cores: 4-8 cores. Memory: 16-32 GB. I / O: High. Medium priority tasks: CPU cores: 2-4 cores. Memory: 8-16 GB. I / O: Medium. Low priority tasks: CPU cores: 1-2 cores. Memory: 2-8 GB. I / O: Low.

[0055] Define the capacity of the resource pool based on the resource bottlenecks that may be encountered in reality. Assume that resources are limited to trigger contention for high-priority tasks: Total number of CPU cores: 200 cores. Total memory capacity: 512 GB. Total I / O bandwidth: 1000 MB / s. The resource pool capacity should be designed to be slightly lower than the sum of the task requirements to ensure that contention occurs. For example: the peak resource demand of a high-priority task is 240 cores, while the actual resources only provide 200 cores. This creates a scenario of insufficient resources to test how the system is scheduled.

[0056] Task submission is simulated according to different scenarios: Peak submission scenario: All tasks are submitted at the same time to simulate sudden loads (such as promotional activities or system stress testing). Batch submission scenario: Tasks are submitted in batches (such as submitting 200 tasks at 10 second intervals). Priority sudden change scenario: Some low-priority tasks are upgraded during execution to test the system's dynamic scheduling capabilities.

[0057] The following typical scheduling algorithms are used to test and compare their performances:

[0058] Optimal fit scheduling algorithm: According to the task resource requirements, find the closest resource block that is sufficient to meet the requirements to avoid resource waste. High-priority tasks may be delayed because low-priority tasks occupy fragmented resources.

[0059] Priority-driven scheduling algorithm: Scheduling is performed strictly according to task priority. High-priority tasks are allocated resources first, and even if low-priority tasks are already running, they will be interrupted or delayed. This ensures that high-priority tasks are completed, but may cause low-priority tasks to not be executed for a long time.

[0060] Weighted fair scheduling algorithm: Set weights for tasks (high-priority tasks have greater weights), and allocate resources in proportion to the weights. While ensuring the execution of high-priority tasks, try to take into account the fairness of medium and low-priority tasks.

[0061] Dynamic priority adjustment algorithm: Dynamically adjust the task priority according to the urgency of the task (such as the remaining deadline, task queue time). Increase the priority of tasks that are about to expire, taking into account both real-time and fairness.

[0062] S2: Run the test according to the designed task queue and resource pool, monitor the resource allocation and task execution of the resource allocation and scheduling system, record the entire process of each task from resource request to allocation and execution completion, and extract the scheduling success rate fluctuation characteristics during the test process and the resolution efficiency characteristics when conflicts occur.

[0063] Start the test according to the task queue and resource pool configuration to observe how the scheduling system handles resource allocation and conflict issues in high concurrency scenarios.

[0064] Start resource pool: Initialize resource pool, including resource parameters such as number of CPU cores, memory capacity, and I / O bandwidth. For example: Total CPU: 200 cores. Total memory: 512 GB. Total I / O bandwidth: 1000 MB / s. Load task queue: Submit the test task queue to the scheduling system. The task contains the following information: Priority (high, medium, low). Resource requirements (CPU, memory, I / O). Duration (run time). Submission time (immediate or delayed submission). Call the scheduling algorithm (such as priority drive, dynamic priority adjustment, etc.) to let the system allocate resources and schedule tasks according to the actual situation of the task queue and resource pool.

[0065] Monitoring resource allocation and task execution includes: task execution status, whether resources are successfully obtained. The actual number of resources allocated (whether the requirements are met). Execution completion time (whether the task deadline is met). Resource pool status: Resource utilization: real-time utilization of CPU, memory, and I / O. Idle resources: whether there are unused idle resources. Resource fragmentation: scattered resources (such as small resource blocks that are not enough to meet task requirements). Record events in which tasks compete for resources (competition between high-priority tasks or resource conflicts between high- and low-priority tasks). Response time for conflict resolution (time from conflict detection to allocation completion). Conflict resolution strategies (such as task interruption, resource reallocation, etc.). Use a log system to record every step of the scheduling system, including resource allocation, task launch, and completion. Real-time visualization tools display resource utilization curves and task status.

[0066] Key data is collected during the test to form the basis for subsequent analysis. Scheduling success rate:

[0067] Record the scheduling status (success / failure) of each task. Calculate the scheduling success rate: Scheduling success rate = number of tasks with successfully allocated resources / total number of tasks; Fluctuation characteristics of scheduling success rate: record the changing trend of scheduling success rate by time period (such as the number of tasks completed per second). Detect fluctuation points (such as a sharp drop in scheduling success rate during task peak). Count the number of resource conflicts that occur during task execution. Record the distribution of task priorities where conflicts occur (such as the conflict ratio of high-priority tasks to low-priority tasks). Conflict resolution time: the average time from detection to conflict resolution. Conflict resolution efficiency = number of conflicts successfully resolved / total time for conflict resolution; success rate and execution effect of different conflict resolution methods (such as resource reservation, dynamic allocation).

[0068] Scheduling success rate fluctuation characteristics: observe the change of success rate over time and identify the peak and valley of fluctuation. Task priority dimension: analyze the scheduling success rates of high, medium and low priority tasks respectively and identify the deviation of priority strategy. Resource dimension: analyze the scheduling failure caused by insufficient specific resources (such as CPU or memory). Whether the success rate of high priority tasks drops significantly in a high concurrency environment. Whether the fluctuation of scheduling success rate is directly related to the utilization or fragmentation of resource pools.

[0069] Conflict resolution efficiency characteristics: Conflict detection time: the response speed of the system after detecting a conflict. Conflict resolution time: the time from detection to completion of resource reallocation. Conflict resolution rate: the proportion of successful resolutions among total conflicts. Whether conflicts between high-priority tasks are resolved first and whether low-priority tasks are reasonably downgraded. Whether the frequency of conflict events increases linearly with the increase in the number of tasks.

[0070] After analyzing the extracted characteristics of the fluctuation of the scheduling success rate, a scheduling success rate fluctuation index is generated. The method for obtaining the scheduling success rate fluctuation index is as follows:

[0071] Assume that the scheduling success rate is recorded by time period during the test, and the time series is: ; In the formula, represents the scheduling success rate in the nth time period (the value range is 0 to 1, or in percentage form from 0% to 100%), and n is the total number of time periods. Calculate the moving average of the scheduling success rate to smooth the data and eliminate short-term noise. The formula is: ; is the moving average in the ith time period, and w is the window size (take the nearest w time periods for mean calculation, typical values are 3, 5, 10, etc.). is the data point of the scheduling success rate within the window. When i < w, only the first i data points are used to calculate the mean.

[0072] Calculate the deviation degree of the scheduling success rate in each time period relative to the moving average , and the formula is: ; In the formula, ; is the deviation degree of the scheduling success rate in the ith time period, which reflects the difference between the current value and the moving average trend. The deviation degrees are integrated into an overall scheduling success rate fluctuation index to reflect the fluctuation amplitude of the entire time series. The formula is: ; In the formula, W is the scheduling success rate fluctuation index.

[0073] A larger scheduling success rate fluctuation index indicates that the scheduling accuracy of the system in a high-concurrency environment is poor. Frequent fluctuations in the success rate mean instability in resource allocation and task scheduling. Especially when high-priority tasks compete for resources, it may lead to unreasonable resource allocation or low efficiency in conflict handling. The real-time requirements of high-priority tasks are difficult to be continuously guaranteed, and in this case, there may be a risk that critical tasks are delayed or failed due to insufficient resources. In addition, a large fluctuation index may also reflect that low-priority tasks over-occupy resources or there is a bottleneck in the system resource pool, resulting in the scheduling system being difficult to stably meet task requirements.

[0074] A smaller scheduling success rate fluctuation index indicates that the scheduling accuracy of the system in a high-concurrency environment is higher, and resource allocation is more stable and reliable. A low fluctuation index indicates that the scheduling system can continuously maintain a high success rate, give priority to meeting the resource requirements of high-priority tasks, and avoid conflicts and waste in resource allocation. In this case, the system shows good resource utilization efficiency and scheduling ability, can effectively handle high-concurrency loads, guarantee the real-time nature of critical tasks, and improve the reliability of overall task execution and the efficiency of enterprise operation.

[0075] After analyzing the extracted conflict resolution efficiency features, a conflict resolution efficiency anomaly index is generated. The method for obtaining the conflict resolution efficiency anomaly index is as follows:

[0076] Assume that the time series of conflict resolution efficiency is: ; is the efficiency of resolving the mth conflict, usually the resolution time (unit: seconds) or the resolution success rate (value range 0 to 1), and m is the total number of data points. Calculate the 25th percentile Q1 and 75th percentile Q3 of the conflict resolution efficiency, as well as the interquartile range IQR, the formula is: Q1=⌊0.25×n⌋th data point; Q3=⌊0.75×n⌋th data point; IQR=Q3−Q1; set the upper and lower bounds of normal data according to the quantile range: lower bound=Q1−1.5×IQR; upper bound=Q3+1.5×IQR; mark data points outside the upper and lower bounds as abnormal points: if <Nether or > upper bound, then is an outlier point, and the ratio of the number of outliers to the total data points m is calculated, that is, the conflict resolution efficiency anomaly index is calculated.

[0077] A large abnormal index of conflict resolution efficiency indicates that the accuracy of resource scheduling in a high-concurrency environment is poor, and the system is unstable when dealing with resource competition for high-priority tasks. This may mean that conflict resolution takes a long time or has a low success rate, resulting in critical tasks being unable to obtain the required resources in a timely manner, thereby delaying or interrupting task execution. In this case, the system's scheduling logic may have problems with improper priority handling, insufficient resource reservation, or ineffective conflict mitigation strategies, making it difficult for high-priority tasks to be effectively guaranteed.

[0078] The small abnormal index of conflict resolution efficiency indicates that the accuracy of resource scheduling in a high-concurrency environment is high, and the system can stably and efficiently solve the problem of resource competition for high-priority tasks. The scheduling system can quickly and accurately allocate resources when conflicts occur, while avoiding excessive delays or failures, providing sufficient execution guarantees for high-priority tasks. This reflects the good performance of the system in priority processing, resource allocation strategy, and dynamic scheduling mechanism, ensuring the real-time and reliability of task execution.

[0079] S3: Based on the extracted scheduling success rate fluctuation characteristics and conflict resolution efficiency characteristics, the accuracy of real-time resource scheduling of high-priority tasks in a high-concurrency environment is evaluated.

[0080] The scheduling success rate fluctuation index and the conflict resolution efficiency anomaly index are converted into comprehensive feature vectors, and the comprehensive feature vectors are used as the input of the machine learning model. The machine learning model uses each group of comprehensive feature vectors to predict the accuracy value label of real-time resource scheduling for high-priority tasks in a high-concurrency environment as the prediction target, and takes minimizing the sum of prediction errors of accuracy value labels for real-time resource scheduling of high-priority tasks in all high-concurrency environments as the training target. The machine learning model is trained until the sum of prediction errors converges, and the model training is stopped. The accuracy value of real-time resource scheduling for high-priority tasks in a high-concurrency environment is determined according to the model output results, wherein the machine learning model is a polynomial regression model.

[0081] S4: Based on the evaluation results, the accuracy of real-time resource scheduling for high-priority tasks in a high-concurrency environment is divided into accurate scheduling, incompletely accurate scheduling, and inaccurate scheduling;

[0082] Compare the acquired accuracy value of real-time resource scheduling of high-priority tasks in a high-concurrency environment with a gradient standard threshold, where the gradient standard threshold includes a first standard threshold and a second standard threshold, and the first standard threshold is less than the second standard threshold, and compare the accuracy value of real-time resource scheduling of high-priority tasks in a high-concurrency environment with the first standard threshold and the second standard threshold respectively;

[0083] If the accuracy value of real-time resource scheduling of high-priority tasks in a high-concurrency environment is greater than the second standard threshold, it means that the accuracy of real-time resource scheduling of high-priority tasks in a high-concurrency environment is high, and a high-accuracy scheduling signal is generated at this time, and it is classified as accuracy scheduling;

[0084] If the accuracy value of real-time resource scheduling of high-priority tasks in a high-concurrency environment is greater than or equal to the first standard threshold and less than or equal to the second standard threshold, it means that the accuracy of real-time resource scheduling of high-priority tasks in a high-concurrency environment is medium, and a medium-accuracy scheduling signal is generated at this time, and it is divided into incomplete accuracy scheduling;

[0085] If the accuracy value of real-time resource scheduling of high-priority tasks in a high-concurrency environment is less than the first standard threshold, it means that the accuracy of real-time resource scheduling of high-priority tasks in a high-concurrency environment is low. At this time, a low-accuracy scheduling signal is generated and classified as inaccurate scheduling.

[0086] S5: For accurate scheduling, the current scheduling algorithm and strategy should be maintained, and the system's resource allocation performance should be continuously monitored; for inaccurate scheduling, the algorithm should be replaced and reconstructed.

[0087] Accurate scheduling shows that the system's resource allocation and task scheduling performance in a high-concurrency environment is stable, and the real-time performance and resource utilization efficiency of high-priority tasks are fully guaranteed. Maintain the current scheduling algorithm and strategy. The current algorithm can meet the system requirements without large-scale adjustments. Ensure that the scheduling system still performs well under continuous high load or changes in task patterns. Prevent performance degradation and ensure that high-priority tasks are completed on time by regularly evaluating the system's resource allocation efficiency and task completion rate. Prevent resource waste or bottlenecks and ensure the system's ability to handle sudden conflicts. Use real-time monitoring tools (such as Prometheus and Grafana) to track resource usage. Generate performance reports regularly to evaluate the performance of the scheduling system under different load conditions. Set thresholds for performance indicators (such as scheduling success rate below 95% and CPU utilization exceeding 90%), and trigger warnings when indicators are abnormal. Use machine learning to predict possible performance bottlenecks, reserve resources in advance, or optimize scheduling strategies.

[0088] Inaccurate scheduling indicates that the system performs poorly in resource allocation and task scheduling in a high-concurrency environment, with frequent resource conflicts, high-priority tasks not being guaranteed, and overall poor performance. Dynamic priority adjustment algorithm: Dynamically adjust the priority according to the urgency of the task (such as the remaining deadline). Resource partitioning and reservation algorithm: Reserve a certain resource pool for high-priority tasks to avoid resource contention. Conflict optimization algorithm: Improve conflict detection and resolution strategies, such as task sharding or dynamic resource expansion. Machine learning scheduling algorithm: Use historical data to train models to predict task requirements and optimize resource allocation decisions. Distributed scheduling system: Use distributed architecture (such as Kubernetes) to improve scheduling parallelism and fault tolerance. Elastic resource expansion mechanism: Combine cloud computing resources to dynamically expand the resource pool to avoid scheduling failures caused by insufficient resources. AI decision module: Intelligent scheduling system based on reinforcement learning or deep learning to optimize task allocation in real time. Verify the performance of the new algorithm under different load conditions to ensure that it is better than the original algorithm. Test indicators include: scheduling success rate, resource utilization, conflict resolution time, etc. Further adjust the algorithm parameters based on the test results to ensure its adaptability in real environments. Compare the scheduling indicators before and after optimization to ensure that the performance is significantly improved. Keep track of the scheduling system performance and record the operation log to discover new problems and make continuous improvements.

[0089] S6: For incomplete accuracy scheduling, predict the degree of abnormality in the accuracy of real-time resource scheduling of high-priority tasks in a high-concurrency environment, and dynamically adjust the resource pool allocation ratio based on the prediction results. When the degree of abnormality is high, switch low-priority tasks to the backup resource pool to release resources for high-priority tasks.

[0090] For incomplete accuracy scheduling, that is, the accuracy value of real-time resource scheduling for high-priority tasks in a high-concurrency environment generated within a fixed time period is greater than or equal to the first standard threshold and less than or equal to the second standard threshold, the accuracy values ​​generated in the subsequent fixed time period that are greater than or equal to the first standard threshold and less than or equal to the second standard threshold are collected, and the corresponding time series is constructed, the mean and standard deviation of the time series are calculated, and after analysis, the abnormal degree of accuracy of real-time resource scheduling for high-priority tasks in a high-concurrency environment is predicted.

[0091] If the mean accuracy value in the data set is greater than or equal to the reference threshold of the mean accuracy value, and the standard deviation of the accuracy value is less than the reference threshold of the standard deviation of the accuracy value, it means that the accuracy value is high and the fluctuation is small (high stability). The system performs well, and although the scheduling accuracy is not optimal, it is maintained within an acceptable range. At this time, the current resource pool allocation ratio should be maintained without large-scale adjustments. Continue to monitor the accuracy value to prevent increased fluctuations.

[0092] If the mean of the accuracy values ​​is greater than or equal to the reference threshold of the mean of the accuracy values, and the standard deviation of the accuracy values ​​is greater than or equal to the reference threshold of the standard deviation of the accuracy values, the accuracy value is high (exceeds the reference mean), but the fluctuation is large (unstable). The system performs well in certain time periods, but there is a risk of significant fluctuations in scheduling accuracy, which may have an intermittent impact on the real-time performance of high-priority tasks. At this time, the resource pool allocation ratio should be adjusted dynamically: more resources should be reserved for high-priority tasks to smooth out fluctuations. Optimize the scheduling algorithm to enhance stability in high-load scenarios. Reduce the resource allocation priority of low-priority tasks to reduce their interference with system stability.

[0093] If the mean accuracy value is less than the reference threshold of the mean accuracy value, and the standard deviation of the accuracy value is greater than or equal to the reference threshold of the standard deviation of the accuracy value, the accuracy value is low (lower than the reference mean) and has large fluctuations (unstable). The overall system performance is poor, the scheduling efficiency is low, and it is difficult to ensure the real-time performance of high-priority tasks. At this time, low-priority tasks should be switched to the backup resource pool, and the main resource pool should be released to prioritize high-priority tasks. Use a dynamic priority adjustment algorithm to promptly increase the priority of tasks approaching the deadline. Dynamically expand cloud resources to relieve high-load pressure.

[0094] If the mean of the accuracy value is less than the reference threshold of the mean of the accuracy value, and the standard deviation of the accuracy value is less than the reference threshold of the standard deviation of the accuracy value, the accuracy value is low (lower than the reference mean), but the fluctuation is small (stable). The overall scheduling performance of the system is poor, but the performance is stable, which may be caused by unreasonable resource pool allocation or improper priority processing, resulting in long-term inefficiency. At this time, the resource allocation ratio of high-priority tasks should be increased, and the occupancy of medium and low-priority tasks should be reduced. Verify whether the scheduling strategy (such as priority drive, resource reservation, etc.) is adapted to the current task load. Gradually optimize system parameters, such as scheduling queue length, task slicing strategy, etc.

[0095] Adjust resource pool allocation ratio in real time based on predicted abnormality level:

[0096] When the abnormality is high (large fluctuation or low mean): switch low-priority tasks to the backup resource pool, and release the main resource pool for high-priority task scheduling. Adjust the ratio of high-, medium-, and low-priority tasks in the resource pool, for example: increase the proportion of high-priority tasks to 60%. Reduce the total proportion of medium- and low-priority tasks to 40%.

[0097] When the abnormality is low (small fluctuation and high mean): maintain the current resource allocation ratio without intervening in the main resource pool. Dynamically optimize the resource pool configuration only when the system load changes.

[0098] In this embodiment, first, by creating test tasks with different priorities and resource requirements, high concurrency scenarios are simulated, and resource pool capacity and scheduling algorithms are defined; then, the test is run and resource allocation is monitored to extract scheduling success rate fluctuation characteristics and conflict resolution efficiency characteristics; then, scheduling accuracy is evaluated based on the characteristics, and scheduling accuracy is divided into accuracy scheduling, incomplete accuracy scheduling, and inaccuracy scheduling. For accuracy scheduling, the existing algorithm is maintained and system performance is continuously monitored; for inaccuracy scheduling, the algorithm is replaced and reconstructed; and for incomplete accuracy scheduling, the resource pool allocation ratio is dynamically adjusted according to the accuracy anomaly prediction results. When the anomaly is high, low-priority tasks are switched to the backup resource pool, and resources are released to prioritize high-priority tasks, thereby improving overall scheduling efficiency and stability.

[0099] Embodiment 2, an enterprise computing resource allocation and scheduling system described in this embodiment includes a scenario configuration module, a feature extraction module, a scheduling accuracy evaluation module, a scheduling classification module, a scheduling strategy execution module and a dynamic adjustment module:

[0100] Scenario configuration module: creates multiple test tasks, including tasks of different priorities and different resource requirements, simulating high-concurrency scenarios in real environments, including several high-priority tasks competing for limited resources, defining resource pool capacity constraints, and setting the scheduling algorithm used in the test scenario;

[0101] Feature extraction module: Run the test according to the designed task queue and resource pool, monitor the resource allocation and task execution of the resource allocation and scheduling system, record the entire process of each task from resource request to allocation and execution completion, and extract the scheduling success rate fluctuation characteristics during the test process and the resolution efficiency characteristics when conflicts occur;

[0102] Scheduling accuracy evaluation module: evaluates the accuracy of real-time resource scheduling of high-priority tasks in a high-concurrency environment based on the extracted scheduling success rate fluctuation characteristics and conflict resolution efficiency characteristics;

[0103] Scheduling classification module: Based on the evaluation results, the accuracy of real-time resource scheduling for high-priority tasks in a high-concurrency environment is divided into accurate scheduling, incompletely accurate scheduling, and inaccurate scheduling;

[0104] Scheduling strategy execution module: For accurate scheduling, the current scheduling algorithm and strategy should be maintained, and the resource allocation performance of the system should be continuously monitored; for inaccurate scheduling, the algorithm should be replaced and reconstructed;

[0105] Dynamic adjustment module: For incomplete accuracy scheduling, the abnormal degree of real-time resource scheduling accuracy of high-priority tasks in a high-concurrency environment is predicted, and the resource pool allocation ratio is dynamically adjusted according to the prediction results. When the abnormal degree is high, low-priority tasks are switched to the backup resource pool to release resources for high-priority tasks.

[0106] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0107] It should be understood that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. A and B can be singular or plural. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship, but it may also indicate an "and / or" relationship. Please refer to the context for specific understanding.

[0108] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0109] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be covered by the protection scope of the present application.

Claims

1. A method for allocating and scheduling computing resources for an enterprise, characterized by: The following steps are involved: S1: Create multiple test tasks, including tasks of different priorities and resource requirements, to simulate high-concurrency scenarios in real environments, including several high-priority tasks competing for limited resources, define resource pool capacity constraints, and set the scheduling algorithm used in the test scenario; S2: Run the test according to the designed task queue and resource pool, monitor the resource allocation and task execution of the resource allocation and scheduling system, record the entire process of each task from resource request to allocation and execution completion, and extract the scheduling success rate fluctuation characteristics during the test process and the resolution efficiency characteristics when conflicts occur; S3: Based on the extracted scheduling success rate fluctuation characteristics and conflict resolution efficiency characteristics, evaluate the accuracy of real-time resource scheduling of high-priority tasks in a high-concurrency environment. Specifically, convert the scheduling success rate fluctuation index and the conflict resolution efficiency abnormality index into a comprehensive feature vector, and use the comprehensive feature vector as the input of the machine learning model. The machine learning model uses each group of comprehensive feature vectors to predict the accuracy value label of real-time resource scheduling of high-priority tasks in a high-concurrency environment as the prediction target, and minimizes the sum of the prediction errors of the accuracy value labels of real-time resource scheduling of high-priority tasks in all high-concurrency environments as the training target. The machine learning model is trained until the sum of the prediction errors converges, and the model training is stopped. The accuracy value of real-time resource scheduling of high-priority tasks in a high-concurrency environment is determined according to the model output results, wherein the machine learning model is a polynomial regression model; S4: Based on the evaluation results, the accuracy of real-time resource scheduling for high-priority tasks in a high-concurrency environment is divided into accurate scheduling, incompletely accurate scheduling, and inaccurate scheduling; S5: For accurate scheduling, the current scheduling algorithm and strategy should be maintained, and the resource allocation performance of the system should be continuously monitored; for inaccurate scheduling, the algorithm should be replaced and reconstructed; S6: For incomplete accuracy scheduling, predict the degree of abnormality in the accuracy of real-time resource scheduling of high-priority tasks in a high-concurrency environment, and dynamically adjust the resource pool allocation ratio based on the prediction results. When the degree of abnormality is high, switch low-priority tasks to the backup resource pool to release resources for high-priority tasks.

2. The method for allocating and scheduling computing resources of an enterprise according to claim 1, characterized in that: In S2, the extracted scheduling success rate fluctuation characteristics are analyzed to generate a scheduling success rate fluctuation index. The method for obtaining the scheduling success rate fluctuation index is: Assume that the scheduling success rate is recorded by time period during the test, and the time series is: ; In the formula, Indicates the scheduling success rate of the nth time period, where n is the number of total time periods; Calculate the moving average of the scheduling success rate, with the formula: ; is the moving average for the i-th time period, w is the window size, are the scheduling success rate data points within the window. When i < w, only the first i data points are used to calculate the mean; Calculate the scheduling success rate for each time period For the moving average Degree of deviation , the formula is: ; In the formula, is the deviation degree of the scheduling success rate for the i-th time period, reflecting the difference between the current value and the moving average trend. Combine the deviation degree into an overall scheduling success rate fluctuation index, reflecting the fluctuation amplitude of the entire time series. The formula is: ; In the formula, W is the scheduling success rate fluctuation index.

3. The method for allocating and scheduling computing resources of an enterprise according to claim 2, characterized in that: In S2, the conflict resolution efficiency anomaly index is generated after analyzing the extracted conflict resolution efficiency features. The method for obtaining the conflict resolution efficiency anomaly index is as follows: Assume that the time series of the conflict resolution efficiency is: ; is the resolution efficiency of the mth conflict, m is the total number of data points, and the 25th percentile Q1 and 75th percentile Q3 of the conflict resolution efficiency, as well as the interquartile range IQR, are calculated using the formula: Q1=⌊0.25×n⌋th data point; Q3 = ⌊0.75×n⌋th data point; IQR=Q3−Q1; set the upper and lower bounds of normal data according to the quantile range: lower bound=Q1−1.5×IQR; upper bound=Q3+1.5×IQR; mark the data points beyond the upper and lower bounds as abnormal points: <Nether or > upper bound, then is an outlier point, and the ratio of the number of outliers to the total data points m is calculated, that is, the conflict resolution efficiency anomaly index is calculated.

4. The method for allocating and scheduling computing resources of an enterprise according to claim 1, characterized in that: In S4, according to the evaluation results, the accuracy of real-time resource scheduling of high-priority tasks in a high-concurrency environment is divided into accurate scheduling, incompletely accurate scheduling, and inaccurate scheduling; Compare the acquired accuracy value of real-time resource scheduling of high-priority tasks in a high-concurrency environment with a gradient standard threshold, where the gradient standard threshold includes a first standard threshold and a second standard threshold, and the first standard threshold is less than the second standard threshold, and compare the accuracy value of real-time resource scheduling of high-priority tasks in a high-concurrency environment with the first standard threshold and the second standard threshold respectively; If the accuracy value of real-time resource scheduling of high-priority tasks in a high-concurrency environment is greater than the second standard threshold, it means that the accuracy of real-time resource scheduling of high-priority tasks in a high-concurrency environment is high, and a high-accuracy scheduling signal is generated at this time, and it is classified as accuracy scheduling; If the accuracy value of real-time resource scheduling of high-priority tasks in a high-concurrency environment is greater than or equal to the first standard threshold and less than or equal to the second standard threshold, it means that the accuracy of real-time resource scheduling of high-priority tasks in a high-concurrency environment is medium, and a medium-accuracy scheduling signal is generated at this time, and it is divided into incomplete accuracy scheduling; If the accuracy value of real-time resource scheduling of high-priority tasks in a high-concurrency environment is less than the first standard threshold, it means that the accuracy of real-time resource scheduling of high-priority tasks in a high-concurrency environment is low. At this time, a low-accuracy scheduling signal is generated and classified as inaccurate scheduling.

5. The method for allocating and scheduling computing resources of an enterprise according to claim 1, characterized in that: In S6, for incomplete accuracy scheduling, the abnormal degree of accuracy of real-time resource scheduling of high-priority tasks in a high concurrency environment is predicted, and the resource pool allocation ratio is dynamically adjusted according to the prediction results. Specifically, for incomplete accuracy scheduling, that is, the accuracy value of real-time resource scheduling of high-priority tasks in a high concurrency environment generated within a fixed time period is greater than or equal to the first standard threshold and less than or equal to the second standard threshold, the accuracy values ​​generated within a subsequent fixed time period that are greater than or equal to the first standard threshold and less than or equal to the second standard threshold are collected, and the corresponding time series is constructed, the mean and standard deviation of the time series are calculated, and after analysis, the abnormal degree of accuracy of real-time resource scheduling of high-priority tasks in a high concurrency environment is predicted.

6. The method for allocating and scheduling computing resources of an enterprise according to claim 5, characterized in that: If the mean accuracy value in the data set is greater than or equal to the reference threshold of the mean accuracy value, and the standard deviation of the accuracy value is less than the reference threshold of the standard deviation of the accuracy value, then the current resource pool allocation ratio should be maintained without large-scale adjustment, and the accuracy value should continue to be monitored to prevent increased fluctuations; If the mean accuracy value is greater than or equal to the reference threshold of the mean accuracy value, and the standard deviation of the accuracy value is greater than or equal to the reference threshold of the standard deviation of the accuracy value, then the resource pool allocation ratio should be dynamically adjusted: more resources should be reserved for high-priority tasks, and the resource allocation priority of low-priority tasks should be reduced; If the mean accuracy value is less than the reference threshold of the mean accuracy value, and the standard deviation of the accuracy value is greater than or equal to the reference threshold of the standard deviation of the accuracy value, the low-priority task should be switched to the backup resource pool, the main resource pool should be released to give priority to the high-priority task, and the dynamic priority adjustment algorithm should be used to timely increase the priority of the deadline task; If the mean accuracy value is less than the reference threshold of the mean accuracy value, and the standard deviation of the accuracy value is less than the reference threshold of the standard deviation of the accuracy value, the resource allocation ratio of high-priority tasks should be increased, and the occupancy of medium and low-priority tasks should be reduced.

7. An enterprise computing resource allocation and scheduling system, used to implement an enterprise computing resource allocation and scheduling method according to any one of claims 1 to 6, characterized in that: It includes scenario configuration module, feature extraction module, scheduling accuracy assessment module, scheduling classification module, scheduling strategy execution module and dynamic adjustment module: Scenario configuration module: creates multiple test tasks, including tasks of different priorities and different resource requirements, simulating high-concurrency scenarios in real environments, including several high-priority tasks competing for limited resources, defining resource pool capacity constraints, and setting the scheduling algorithm used in the test scenario; Feature extraction module: Run the test according to the designed task queue and resource pool, monitor the resource allocation and task execution of the resource allocation and scheduling system, record the entire process of each task from resource request to allocation and execution completion, and extract the scheduling success rate fluctuation characteristics during the test process and the resolution efficiency characteristics when conflicts occur; Scheduling accuracy evaluation module: evaluates the accuracy of real-time resource scheduling of high-priority tasks in a high-concurrency environment based on the extracted scheduling success rate fluctuation characteristics and conflict resolution efficiency characteristics; Scheduling classification module: Based on the evaluation results, the accuracy of real-time resource scheduling for high-priority tasks in a high-concurrency environment is divided into accurate scheduling, incompletely accurate scheduling, and inaccurate scheduling; Scheduling strategy execution module: For accurate scheduling, the current scheduling algorithm and strategy should be maintained, and the resource allocation performance of the system should be continuously monitored; for inaccurate scheduling, the algorithm should be replaced and reconstructed; Dynamic adjustment module: For incomplete accuracy scheduling, the abnormal degree of real-time resource scheduling accuracy of high-priority tasks in a high-concurrency environment is predicted, and the resource pool allocation ratio is dynamically adjusted according to the prediction results. When the abnormal degree is high, low-priority tasks are switched to the backup resource pool to release resources for high-priority tasks.

Images