Resource allocation and management system based on dynamic scheduling algorithm

Abstract

The invention discloses a resource allocation and management system based on a dynamic scheduling algorithm, which relates to the field of data center resource allocation and management, and specifically comprises a data center resource monitoring module, a running state feature extraction module, a resource utilization evaluation module and a resource allocation management module, wherein the data center resource monitoring module monitors, collects and analyzes operation state data of a data center in real time to ensure data accuracy and classify the data according to tasks, and the operation state feature extraction module is used for screening and extracting resource features in the operation state data and calculating the resource utilization rate by integrating the resource feature data. The resource utilization evaluation module is used for evaluating the resource utilization efficiency when the data center executes tasks and judging and identifying to-be-allocated tasks through resource utilization evaluation indexes, and the resource allocation management module is used for establishing a simulation model by collecting information of the data center and dynamically allocating resources through an ant colony algorithm. And the task can be efficiently executed within the limited time.

Description

Technical Field

[0001] The present invention belongs to the technical field of data center resource allocation and management, and in particular relates to a resource allocation and management system based on a dynamic scheduling algorithm. Background Art

[0002] As the scale of data centers expands, existing data centers face the serious problem of uneven resource allocation when performing different tasks, and it is difficult to meet new demands. First, in order to meet diverse application requests and improve the performance of data centers, data centers need to be able to flexibly allocate various resources, such as CPU, memory, storage, and network bandwidth. However, in actual operations, due to differences in task characteristics and resource requirements, some tasks often receive too many resources while others are under-resourced. This problem of uneven resource allocation seriously affects the overall efficiency and performance of data centers.

[0003] Secondly, uneven resource allocation may also lead to resource bottlenecks within the data center. When certain tasks cannot be executed efficiently due to insufficient resources, they may become bottlenecks for the entire system, slowing down the execution of other tasks and even causing a decline in the service quality of the entire data center. In addition, due to uneven resource allocation, some high-priority or urgent tasks may not be able to obtain the required resources in a timely manner, thus affecting the normal operation of the business and customer satisfaction.

[0004] Therefore, designing an efficient resource allocation management method to solve the problem of uneven resource allocation when data centers perform different tasks and improve the overall efficiency and performance of data centers is one of the key issues that need to be urgently addressed in the current data center field. Summary of the invention

[0005] The purpose of the present invention is to provide a resource allocation and management system based on a dynamic scheduling algorithm, which is used to solve the technical problem of uneven resource allocation when a data center executes different tasks in the prior art.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] A resource allocation and management system based on a dynamic scheduling algorithm comprises: a data center resource monitoring module, an operation status feature extraction module, a resource utilization evaluation module and a resource allocation management module.

[0008] Among them, the data center resource monitoring module collects the physical layout, equipment configuration, network structure and task processing flow information of the data center, and determines the changes in physical parameters during the execution of tasks by the data center, where the physical parameters refer to: the effective load capacitance, voltage threshold and operating frequency of the data center equipment. The historical data of the data center is collected and analyzed through the data center's log system, and the operating status data of the data center when executing tasks is monitored and determined in real time. The operating status data includes CPU usage, memory usage, disk usage, and network bandwidth usage indicators. Duplicate, invalid or abnormal data is removed to ensure the accuracy and reliability of the operating status data. The collected operating status data is converted into a unified format, classified according to the tasks performed by the data center, and different operating status data when the data center performs the same type of tasks in the historical data are determined.

[0009] The operation status feature extraction module screens the operation status data of the data center completed within the time limit required by the task, and extracts the resource feature data in the operation status data. The resource feature data includes: the resource utilization rate, duration, throughput, and error rate of the data center during the execution of a task. The time point when the data center starts to execute task x is recorded as the starting time point, and the time point when the data center completes task x is recorded as the ending time point. The time period between the starting time point and the ending time point is divided into time periods every T time length, and the operation status data of the time period including the ending time point is not considered. The resource utilization rate is obtained by combining the CPU utilization rate, memory utilization rate, disk utilization rate, and network bandwidth utilization rate allocated to task x in different time periods. The specific calculation formula is as follows:

[0010]

[0011] Wherein, x represents task x executed by the data center, h(x, t) represents the resource usage rate of the data center allocated to task x in the tth time period, cp(x, t) represents the average CPU usage rate of the data center allocated to task x in the tth time period, ne(x, t) represents the average memory occupancy rate of the data center allocated to task x in the tth time period, ci(x, t) represents the average disk usage rate of the data center allocated to task x in the tth time period, wl(x, t) represents the average network bandwidth occupancy rate of the data center allocated to task x in the tth time period, w1 represents the weight coefficient of the average CPU usage rate, w2 represents the weight coefficient of the average memory usage rate, w3 represents the weight coefficient of the average disk usage rate, and w4 represents the weight coefficient of the average network bandwidth usage rate.

[0012] The resource utilization evaluation module determines the average power output caused by the use of CPU, memory, disk, and network bandwidth in the process of executing a task in the data center, which is recorded as the average power consumption. By integrating the average power consumption, delay, throughput, and error rate when the data center performs different tasks, the resource utilization evaluation index is set to evaluate the resource utilization efficiency of the data center when performing different tasks. The specific formula of the resource utilization evaluation index is as follows:

[0013]

[0014] Wherein, R(x, t) represents the resource utilization evaluation index of the data center executing task x in the tth time period, td(x, t) represents the throughput of the data center executing task x in the tth time period, sy(x, t) represents the duration of the data center executing task x in the tth time period, cw(x, t) represents the error rate of the data center executing task x in the tth time period, p(x, t) represents the average power consumption of the data center executing task x in the tth time period, T represents the length of a time period when the data center starts to execute task x, and e is the base of the logarithm of natural numbers.

[0015] The calculation formula for average power consumption is determined based on effective load capacitance, supply voltage, and operating frequency. The specific calculation formula is as follows:

[0016] p(x, t) = dr(x, t) * gy 2 (x, t)*yx(x, t);

[0017] Wherein, p(x, t) represents the average power consumption of the data center executing task x in the t-th time period, dr(x, t) represents the effective load capacitance of the data center executing task x in the t-th time period, gy(x, t) represents the supply voltage of the data center executing task x in the t-th time period, and the supply voltage is determined by the time extension length and the voltage threshold, and yx(x, t) represents the operating frequency of the data center executing task x in the t-th time period;

[0018] Furthermore, the supply voltage is determined by the duration of the data center executing task x and the voltage threshold of the data center executing task x. The specific expression formula of the supply voltage in the t-th time period is as follows:

[0019]

[0020] Wherein, sy(x, t) represents the duration of the data center executing task x in the t-th time period, gy(x, t) represents the supply voltage of the data center executing task x in the t-th time period, g is the correction parameter, f represents the circuit saturation rate, and V represents the voltage threshold.

[0021] Through in-depth analysis and mining of the past historical data of the data center, a reasonable threshold U of the resource utilization evaluation index is set. If the product of the resource utilization rate and the resource utilization evaluation index in each time period from the start of the data center executing task x to the current moment is less than the threshold U, it is judged that the data center has poor resource utilization in the process of executing task x, and task x is recorded as a task to be assigned. On the contrary, if the product of the resource utilization rate and the resource utilization evaluation index in each time period from the start of the data center executing task x to the current moment is greater than or equal to the threshold U, it is judged that the data center has good resource utilization in the process of executing task x, and there is no need to adjust the resources of the current data center for task x.

[0022] The resource allocation management module collects the physical layout, equipment configuration, network structure and task processing flow information of the data center, uses the data center simulator to establish a data center simulation model to simulate the operation of the data center during the execution of different tasks, and collects the operation status data of the data center simulation model in real time when executing different tasks through the sensors and log system tools set in the data center simulation model, ensuring that the collected data covers all key components of the data center simulation model, including physical servers, storage devices, network devices, and virtual machines. The operation status data of the data center executing the task to be assigned when the average resource utilization evaluation index is less than the threshold U is determined, and the operation status data of the task to be assigned is input into the data center simulation model. The scheduling algorithm based on the optimization idea of ​​the ant colony algorithm is used to generate resource allocation combinations for N rounds of tasks to be assigned;

[0023] Determine the set of tasks to be assigned, expressed as {x1, ..., xi, ..., xn}, where n represents a total of n tasks, xi represents the i-th task, and determine the time limit of each task xi as di. According to the actual operation of the data center, dynamically allocate resources for each task to be assigned through a scheduling algorithm based on the optimization idea of ​​the ant colony algorithm. Set the number of ant sets, pheromone volatility coefficient, pheromone initial value, pheromone enhancement coefficient and maximum number of iterations. Each ant obtains different resource utilization rates by determining different CPU utilization rates, memory utilization rates, disk utilization rates and network bandwidth utilization rates. {a1, ..., aj, ..., aN} represents the ant set, aj is recorded as the j-th ant, the CPU usage, memory occupancy, disk usage and network bandwidth occupancy set determined by each ant is recorded as the allocation set, h[aj] is used to represent the allocation set represented by the j-th ant, H[xi] is used to represent the range of the allocation set that can be mobilized in task xi, each ant mainly adopts the probabilistic transfer rule, and n(xi, aj) is used to represent the allocation set represented by ant aj to task xi, each time the allocation combination with the largest probability is selected as the updated allocation combination, and the algorithm is repeatedly run until the maximum number of iterations is reached.

[0024] Among them, the probability transfer rule is specifically expressed by the following method:

[0025]

[0026] Among them, h[aj]→h[aj′] means updating the allocation set represented by the jth ant to the allocation set represented by the j′th ant. represents the state transition probability of the assignment set of task xi from h[aj] to h[aj′], The pheromone representing the assignment set of task xi from h[aj] to h[aj′], It represents the heuristic value calculated by the heuristic function after the allocation set of task xi changes from h[aj] to h[aj′], (u, v) represents any two different allocation sets within the range of the allocation set that can be mobilized in task xi, α represents the weight coefficient of pheromone, and β represents the weight coefficient of the heuristic value.

[0027] Determine the completion deadline of task xi, take maximizing the resource utilization evaluation index of task xi as the objective function, output the allocation set, and use the data center simulation model to determine whether the completion time of task xi meets the deadline after allocating the resources of task xi with the allocation set. When the task deadline is met, output the allocation set. When the task deadline is not met, remove the allocation set from the range of the movable allocation set in task xi, and then find a new allocation set through the objective function until the new allocation set meets the deadline and outputs the new allocation set.

[0028] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0029] 1. The present invention realizes refined task scheduling and resource optimization of the data center through comprehensive data center simulation and monitoring, and through data center operation status data, establishes a data center simulation model based on physical parameters during the operation of the data center, realizes accurate verification of multi-source allocation sets, helps to quickly adjust the data center allocation set, and realizes comprehensive optimization and improvement of data center resource allocation management;

[0030] 2. The present invention sets a resource utilization evaluation index, takes the resource utilization evaluation index of the maximized task as the objective function, verifies the allocation combination that meets the objective function in the data center simulation model, and screens out the allocation combination that maximizes the objective function based on the task completion time limit, which helps to improve the feasibility and effectiveness of task scheduling;

[0031] 3. The present invention determines the set of tasks to be assigned and uses a scheduling algorithm based on the optimization concept of the ant colony algorithm to dynamically and intelligently generate a resource allocation set for N rounds of tasks to be assigned. Through the probabilistic transfer rule, the ants can comprehensively consider information hormones and heuristic values ​​when selecting the allocation set, thereby selecting the optimal allocation set, thereby improving the precision of task scheduling and the efficiency of resource optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0033] Figure 1 A program diagram of a resource allocation and management system based on a dynamic scheduling algorithm is shown;

[0034] Figure 2 A step diagram of a resource allocation and management method based on a dynamic scheduling algorithm is shown;

[0035] Figure 3 A step diagram of a method for determining resource utilization evaluation indicators is shown. DETAILED DESCRIPTION

[0036] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only 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.

[0037] like Figure 1 , Figure 2 , Figure 3 As shown, a resource allocation and management system based on a dynamic scheduling algorithm specifically includes:

[0038] The data center resource monitoring module collects the data center's physical layout, equipment configuration, network structure, and task processing flow information, and determines the changes in physical parameters during the data center's task execution. The physical parameters refer to: the effective load capacitance, voltage threshold, and operating frequency of the data center equipment. The data center's historical data is collected and analyzed through the data center's log system. The operating status data of the data center when executing tasks is monitored and determined in real time. The operating status data includes CPU usage, memory usage, disk usage, and network bandwidth usage indicators. Duplicate, invalid, or abnormal data is removed to ensure the accuracy and reliability of the operating status data. The collected operating status data is converted into a unified format, classified according to the tasks performed by the data center, and different operating status data when the data center performs the same type of tasks in the historical data is determined.

[0039] The operation status feature extraction module screens the operation status data of the data center completed within the time limit required by the task, and extracts the resource feature data in the operation status data. The resource feature data includes: the resource utilization rate, duration, throughput, and error rate of the data center during the execution of a task. The time point when the data center starts to execute task x is recorded as the starting time point, and the time point when the data center completes task x is recorded as the ending time point. The time period between the starting time point and the ending time point is divided into time periods every T time length, and the operation status data of the time period including the ending time point is not considered. The resource utilization rate is obtained by combining the CPU utilization rate, memory utilization rate, disk utilization rate, and network bandwidth utilization rate allocated to task x in different time periods. The specific calculation formula is as follows:

[0040]

[0041] Wherein, x represents task x executed by the data center, h(x, t) represents the resource usage rate of the data center allocated to task x in the tth time period, cp(x, t) represents the average CPU usage rate of the data center allocated to task x in the tth time period, ne(x, t) represents the average memory occupancy rate of the data center allocated to task x in the tth time period, ci(x, t) represents the average disk usage rate of the data center allocated to task x in the tth time period, wl(x, t) represents the average network bandwidth occupancy rate of the data center allocated to task x in the tth time period, w1 represents the weight coefficient of the average CPU usage rate, w2 represents the weight coefficient of the average memory usage rate, w3 represents the weight coefficient of the average disk usage rate, and w4 represents the weight coefficient of the average network bandwidth usage rate.

[0042] The resource utilization evaluation module determines the average power output caused by the use of CPU, memory, disk, and network bandwidth in the process of executing a task in the data center, which is recorded as the average power consumption. By integrating the average power consumption, delay, throughput, and error rate when the data center performs different tasks, the resource utilization evaluation index is set to evaluate the resource utilization efficiency of the data center when performing different tasks. The specific formula of the resource utilization evaluation index is as follows:

[0043]

[0044] Among them, R(x, t) represents the resource utilization evaluation index of the data center executing task x in the tth time period, td(x, t) represents the throughput of the data center executing task x in the tth time period, sy(x, t) represents the duration of the data center executing task x in the tth time period, cw(x, t) represents the error rate of the data center executing task x in the tth time period, p(x, t) represents the average power consumption of the data center executing task x in the tth time period, T represents the length of a time period when the data center starts to execute task x, and e is the base of the logarithm of natural numbers. In this embodiment, e=2.7.

[0045] The calculation formula for average power consumption is determined based on effective load capacitance, supply voltage, and operating frequency. The specific calculation formula is as follows:

[0046] p(x, t) = dr(x, t) * gy 2 (x, t)*yx(x, t);

[0047] Wherein, p(x, t) represents the average power consumption of the data center executing task x in the t-th time period, dr(x, t) represents the effective load capacitance of the data center executing task x in the t-th time period, gy(x, t) represents the supply voltage of the data center executing task x in the t-th time period, and yx(x, t) represents the operating frequency of the data center executing task x in the t-th time period;

[0048] Furthermore, the supply voltage is determined by the duration of the data center executing task x and the voltage threshold of the data center executing task x. The specific expression formula of the supply voltage in the t-th time period is as follows:

[0049]

[0050] Wherein, sy(x, t) represents the duration of the data center executing task x in the t-th time period, gy(x, t) represents the supply voltage of the data center executing task x in the t-th time period, g is a correction parameter, and g=0.5 is set in this embodiment, f represents the circuit saturation rate, and f=2 in this embodiment, and V represents the voltage threshold.

[0051] Through in-depth analysis and mining of the data center's past historical data, a reasonable threshold U of the resource utilization evaluation index is set. If the product of the resource utilization rate and the resource utilization evaluation index in each time period from the start of the data center's execution of task x to the current moment is less than the threshold U, it is judged that the data center has poor resource utilization in the process of executing task x, and task x is recorded as a task to be assigned. Conversely, if the product of the resource utilization rate and the resource utilization evaluation index in each time period from the start of the data center's execution of task x to the current moment is greater than or equal to the threshold U, it is judged that the data center has good resource utilization in the process of executing task x, and there is no need to adjust the resources of the current data center for task x.

[0052] The resource allocation management module collects the physical layout, equipment configuration, network structure and task processing flow information of the data center, uses the data center simulator to establish a data center simulation model to simulate the operation of the data center during the execution of different tasks, and collects the operation status data of the data center simulation model in real time when executing different tasks through the sensors and log system tools set in the data center simulation model, ensuring that the collected data covers all key components of the data center simulation model, including physical servers, storage devices, network devices, and virtual machines. The operation status data of the data center executing the task to be assigned when the average resource utilization evaluation index is less than the threshold U is determined, and the operation status data of the task to be assigned is input into the data center simulation model. The scheduling algorithm based on the optimization idea of ​​the ant colony algorithm is used to generate resource allocation combinations for N rounds of tasks to be assigned;

[0053] Determine the set of tasks to be assigned, expressed as {x1, ..., xi, ..., xn}, where n represents a total of n tasks, xi represents the i-th task, and determine the time limit of each task xi as di. According to the actual operation of the data center, dynamically allocate resources for each task to be assigned through a scheduling algorithm based on the optimization idea of ​​the ant colony algorithm. Set the number of ant sets, pheromone volatility coefficient, pheromone initial value, pheromone enhancement coefficient and maximum number of iterations. Each ant obtains different resource utilization rates by determining different CPU utilization rates, memory utilization rates, disk utilization rates and network bandwidth utilization rates. {a1, ..., aj, ..., aN} represents the ant set, aj is recorded as the j-th ant, the CPU usage, memory occupancy, disk usage and network bandwidth occupancy set determined by each ant is recorded as the allocation set, h[aj] is used to represent the allocation set represented by the j-th ant, H[xi] is used to represent the range of the allocation set that can be mobilized in task xi, each ant mainly adopts the probabilistic transfer rule, and n(xi, aj) is used to represent the allocation set represented by ant aj to task xi, each time the allocation combination with the largest probability is selected as the updated allocation combination, and the algorithm is repeatedly run until the maximum number of iterations is reached.

[0054] Among them, the probability transfer rule is specifically expressed by the following method:

[0055]

[0056] Among them, h[aj]→h[aj′] means updating the allocation set represented by the jth ant to the allocation set represented by the j′th ant. represents the state transition probability of the assignment set of task xi from h[aj] to h[aj′], The pheromone representing the assignment set of task xi from h[aj] to h[aj′], It represents the heuristic value calculated by the heuristic function after the allocation set of task xi changes from h[aj] to h[aj′], (u, v) represents any two different allocation sets within the range of the allocation set that can be mobilized in task xi, α represents the weight coefficient of pheromone, and β represents the weight coefficient of the heuristic value.

[0057] Determine the completion deadline of task xi, take maximizing the resource utilization evaluation index of task xi as the objective function, output the allocation set, and use the data center simulation model to determine whether the completion time of task xi meets the deadline after allocating the resources of task xi with the allocation set. When the task deadline is met, output the allocation set. When the task deadline is not met, remove the allocation set from the range of the movable allocation set in task xi, and then find a new allocation set through the objective function until the new allocation set meets the deadline and outputs the new allocation set.

[0058] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

[0059] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only specific implementation methods. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A resource allocation and management system based on a dynamic scheduling algorithm, characterized in that: include: The data center resource monitoring module monitors and collects the operating status data and physical parameters of the data center when performing different tasks in real time, and classifies the operating status data and physical parameters according to the task type; The operation status feature extraction module is used to filter and extract resource feature data from the operation status data, and calculate the resource utilization rate based on the resource feature data; The resource utilization evaluation module integrates the operation status data and physical parameters of the data center when executing different tasks to set resource utilization evaluation indicators, and combines the resource utilization rate and resource utilization evaluation indicators when the data center executes tasks to judge and identify the tasks to be assigned; The resource allocation management module uses the ant colony algorithm to analyze the tasks to be assigned, adopts the probabilistic transfer rule to update the allocation set of the tasks to be assigned, and outputs the allocation set that meets the resource allocation and management requirements of the tasks to be assigned.

2. A resource allocation and management system based on a dynamic scheduling algorithm according to claim 1, characterized in that: Real-time monitoring and collection of data center operation status data and physical parameters, and classification of operation status data and physical parameters according to task type, including: Real-time monitoring and determination of the operating status data and physical parameters of the data center when performing different tasks. The operating status data includes CPU usage, memory usage, disk usage, and network bandwidth usage indicators. Physical parameters refer to the effective load capacitance, voltage threshold, and operating frequency of the data center equipment, which are classified according to the type of tasks performed by the data center.

3. A resource allocation and management system based on a dynamic scheduling algorithm according to claim 1, characterized in that: The resource utilization rate is calculated by comprehensive resource characteristic data. The specific method is as follows: Using the formula represents resource utilization, where x represents task x executed by the data center, h(x, t) represents the resource utilization allocated to task x by the data center in the tth time period, cp(x, t) represents the average CPU utilization allocated to task x by the data center in the tth time period, ne(x, t) represents the average memory occupancy allocated to task x by the data center in the tth time period, ci(x, t) represents the average disk utilization allocated to task x by the data center in the tth time period, wl(x, t) represents the average network bandwidth utilization allocated to task x by the data center in the tth time period, w1 represents the weight coefficient of the average CPU utilization, w2 represents the weight coefficient of the average memory utilization, w3 represents the weight coefficient of the average disk utilization, and w4 represents the weight coefficient of the average network bandwidth utilization.

4. A resource allocation and management system based on a dynamic scheduling algorithm according to claim 1, characterized in that: The resource utilization evaluation index is set by integrating the operation status data and physical parameters of the data center when performing different tasks. The specific method is as follows: Using the formula represents the resource utilization evaluation index, where R(x, t) represents the resource utilization evaluation index of the data center executing task x in the tth time period, td(x, t) represents the throughput of the data center executing task x in the tth time period, sy(x, t) represents the duration of the data center executing task x in the tth time period, cw(x, t) represents the error rate of the data center executing task x in the tth time period, p(x, t) represents the average power consumption of the data center executing task x in the tth time period, and the average power consumption is determined by the effective load capacitance, voltage threshold and operating frequency, T represents the length of a time period when the data center starts to execute task x, and e is the base of the logarithm of the natural number.

5. A resource allocation and management system based on a dynamic scheduling algorithm according to claim 4, characterized in that: The average power consumption is determined by the effective load capacitance, voltage threshold and operating frequency as follows: Using the formula p(x, t) = dr(x, t)*gy 2 (x, t)*yx(x, t) represents the average power consumption, where p(x, t) represents the average power consumption of the data center executing task x in the t-th time period, dr(x, t) represents the effective load capacitance of the data center executing task x in the t-th time period, gy(x, t) represents the supply voltage of the data center executing task x in the t-th time period, and yx(x, t) represents the operating frequency of the data center executing task x in the t-th time period. The supply voltage is determined by the voltage threshold and the duration.

6. A resource allocation and management system based on a dynamic scheduling algorithm according to claim 5, characterized in that: The supply voltage is determined by the voltage threshold and the time extension. The specific method is as follows: Using the formula It represents the relationship between voltage threshold, time extension and supply voltage, where sy(x, t) represents the time extension of the data center executing task x in the t-th time period, gy(x, t) represents the supply voltage of the data center executing task x in the t-th time period, g is the correction parameter, f represents the circuit saturation rate, and V represents the voltage threshold.

7. A resource allocation and management system based on a dynamic scheduling algorithm according to claim 1, characterized in that: Combine the resource utilization rate and resource utilization evaluation index when the data center executes tasks to determine and identify the tasks to be assigned. The specific method is as follows: Set a threshold U for the resource evaluation index. If the product of the resource utilization rate and the resource utilization evaluation index is less than the threshold U when the data center executes the task in each time period from the time when the data center executes the task x to the current moment, the task is recorded as a task to be assigned. Conversely, if the product of the resource utilization rate and the resource utilization evaluation index in each time period is greater than or equal to the threshold U, the resource allocation and management of the task is judged to be normal.

8. The resource allocation and management system based on dynamic scheduling algorithm according to claim 1, characterized in that: Ant colony algorithm is used to analyze the tasks to be assigned. The specific method is as follows: Determine the set of tasks to be assigned, expressed as {x1, ..., xi, ..., xn}, where n represents a total of n tasks, xi represents the i-th task, and the time limit of each task xi is determined to be di. According to the actual operation of the data center, dynamic resource allocation is performed for each task to be assigned through a scheduling algorithm based on the optimization idea of ​​the ant colony algorithm. The number of ant sets, pheromone volatility coefficient, pheromone initial value, pheromone enhancement coefficient and maximum number of iterations are set. Each ant determines different CPU usage, memory occupancy, disk usage and network bandwidth occupancy, which are recorded as allocation sets. {a1, ..., aj, ..., aN} represents the ant set, aj is recorded as the j-th ant, h[aj] is used to represent the allocation set represented by the j-th ant, H[xi] is used to represent the range of the allocation set that can be mobilized in task xi, and each ant mainly adopts the probability transfer rule. n(xi, aj) is used to represent the allocation set represented by ant aj to task xi. Each time, the allocation combination with the largest probability is selected as the updated allocation combination, and the algorithm is repeatedly run until the maximum number of iterations is reached.

9. The resource allocation and management system based on dynamic scheduling algorithm according to claim 1, characterized in that: The probability transfer rule is used to update the allocation set of the tasks to be assigned. The probability transfer rule is specifically expressed by the following formula: Among them, h[aj]→h[aj′] means updating the allocation set of the j-th ant representative to the allocation set of the j′-th ant representative. represents the state transition probability of the assignment set of task xi from h[aj] to h[aj′], The pheromone representing the assignment set of task xi from h[aj] to h[aj′], It represents the heuristic value calculated by the heuristic function after the allocation set of task xi changes from h[aj] to h[aj′], (u, v) represents any two different allocation sets within the range of the allocation set that can be mobilized in task xi, α represents the weight coefficient of pheromone, and β represents the weight coefficient of the heuristic value.

10. The resource allocation and management system based on dynamic scheduling algorithm according to claim 1, characterized in that: Output the allocation set that meets the resource allocation and management requirements of the tasks to be assigned. The specific method is: Determine the completion deadline of task xi, take maximizing the resource utilization evaluation index of task xi as the objective function, output the allocation set, and use the data center simulation model to determine whether the completion time of task xi meets the deadline after allocating the resources of task xi with the allocation set. When the task deadline is met, output the allocation set. When the task deadline is not met, remove the allocation set from the range of the movable allocation set in task xi, and then find a new allocation set through the objective function until the new allocation set meets the deadline and outputs the new allocation set.

Images