A resource scheduling method and system based on data center

By analyzing the remaining capacity and job requirements of each queue in the data center, and judging and moving jobs in the queue to make room for new jobs, the problem of resource waste is solved and resource utilization and job processing efficiency is improved.

CN118708326BActive Publication Date: 2024-11-22SHANDONG CHENHUA TECH INFORMATION CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411210415.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-30
Publication Date
2024-11-22
Estimated Expiration
2044-08-30

AI Technical Summary

Technical Problem

In the prior art, when the submitted job capacity is greater than the capacity that any queue can accommodate, job waiting will result in waste of resources, and the remaining capacity in the queue will not be effectively utilized.

Method used

By obtaining the remaining capacity of each queue and the job requirement capacity, determine whether there is a accommodated queue. If it does not exist, then move the jobs of other queues to make room, select the appropriate queue to place new jobs, and reduce resource waste.

Benefits of technology

When adding new jobs, analyze the remaining capacity of each queue, try to move the jobs to make room, reduce resource waste, improve resource utilization, and ensure that the jobs can enter the queue for processing appropriately.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118708326B_ABST
    Figure CN118708326B_ABST
Patent Text Reader

Abstract

The present application relates to a resource scheduling method and system based on a data center, and to the field of cloud computing technology. The method includes obtaining the job demand capacity of the current newly added job, the queue consumption capacity of each queue, and the remaining capacity of the queue; and accommodating queues and fully loaded queues; defining alternative queues when there are no accommodating queues, and determining the required empty capacity according to the alternative queues; defining each job in the alternative queue as an alternative moving job, and obtaining the monomer moving capacity; determining the moving job combination and the combined moving capacity according to the random combination of each alternative moving job, and defining the required combination and the required moving job; determining the monomer moving order according to the required moving job, and selecting the placement queue of the required moving job according to the monomer moving order, and defining it as an optional combination; selecting a unique operation combination in the optional combination, and adjusting the job according to the operation combination. The present application has the effect of reducing the occurrence of resource waste.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of cloud computing technology, and in particular to a data center-based resource scheduling method and system. Background Art

[0002] With the rapid development of cloud computing technology, data center operating systems are gradually playing an increasingly important role in social production and life. They can uniformly manage various resources in the data center and perform resource scheduling. The various resources in the data center include computing resources, network resources, storage resources, etc.

[0003] In the related technologies, the allocation of computing resources is usually done by FIFO strategy, shortest job first strategy, capacity scheduling strategy, etc. Among them, for the capacity scheduling strategy, multiple job queues will be created first. When the user submits a new job, the scheduling strategy will select a suitable queue for the new job based on the job load and resource capacity of each queue.

[0004] Regarding the above-mentioned related technologies, the inventor believes that when the capacity of the submitted jobs is greater than the capacity of any queue, new jobs will need to wait outside the queue, while each queue still has some capacity available for use, resulting in a waste of resources. There is still room for improvement. Summary of the invention

[0005] In order to reduce the occurrence of resource waste, the present application provides a data center-based resource scheduling method and system.

[0006] In a first aspect, the present application provides a data center-based resource scheduling method, which adopts the following technical solution:

[0007] A resource scheduling method based on a data center, comprising:

[0008] Obtain the job demand capacity of the current newly added job, the queue consumption capacity of each queue, and the remaining capacity of the queue;

[0009] The queues whose remaining capacity is not less than the job demand capacity are defined as accommodating queues, and the remaining queues are defined as fully loaded queues;

[0010] Determine whether there is a queue that can be accommodated;

[0011] If there are queues that can be accommodated, a unique queue is selected from each queue that can be accommodated, and the current newly added job is added to the queue;

[0012] If there is no queue that can accommodate, the queue with the largest remaining capacity is defined as the candidate queue, and the difference between the remaining capacity of the candidate queue and the job demand capacity is calculated to determine the required vacant capacity;

[0013] Define each job in the candidate queue as a candidate move job, and obtain the individual move capacity of each candidate move job;

[0014] Perform any number of random combinations according to the alternative moving operations to determine the moving operation combination, and perform sum calculation according to the individual moving capacities of the alternative moving operations in the moving operation combination to determine the combined moving capacity, and define the moving operation combination whose combined moving capacity is not less than the required moving capacity as the required combination, and define the alternative moving operations in the required combination as the required moving operations;

[0015] According to the monomer moving capacity of the required moving operation, the monomer moving order is determined by sorting from large to small, and the placement queue of the required moving operation is selected from the remaining full-load queues in turn according to the monomer moving order, and when each required moving operation selects a placement queue, the combination that meets the requirements is defined as an optional combination;

[0016] A unique operation combination is selected from the optional combinations, and each required moving operation is controlled to move to a corresponding placement queue according to the operation combination, and the current newly added operation is added to the optional queue.

[0017] Optionally, the step of selecting a unique queuing queue from each accommodating queue includes:

[0018] Get the waiting job capacity of the jobs waiting in each queue;

[0019] Determine the waiting operation capacity with the smallest value according to a preset sorting rule, and define the waiting operation capacity as the benchmark operation capacity;

[0020] The capacity after adding is determined by calculating the difference between the remaining capacity of each queue that can be accommodated and the required capacity of the job;

[0021] Determine the first selection parameter corresponding to the added capacity according to the preset first selection matching relationship;

[0022] Perform difference calculation based on the added capacity and the benchmark operation capacity to determine the operation difference capacity, and determine the second selection parameter corresponding to the operation difference capacity based on a preset second selection matching relationship;

[0023] A calculation is performed according to the first selection parameter and the second selection parameter to determine a comprehensive selection parameter, and the comprehensive selection parameter with the largest value is determined according to the sorting rule, and the accommodating queue corresponding to the comprehensive selection parameter is determined as the queuing queue.

[0024] Optionally, after the comprehensive selection parameters are determined, the resource scheduling method based on the data center further includes:

[0025] Determine whether there are at least two queues with consistent comprehensive selection parameters and the largest accommodating queues;

[0026] If there are no at least two queues with the same comprehensive selection parameters and the largest accommodating queue, the queue is determined according to the only comprehensive selection parameter;

[0027] If there are at least two maximum accommodating queues with the same comprehensive selection parameters, the consumption ratio is determined by calculating the queue consumption capacity and the queue remaining capacity of each accommodating queue;

[0028] The number of jobs waiting on each accommodating queue is counted to determine the number of queue waiting;

[0029] The priority factor is determined by calculating the consumption ratio and the number of queues waiting;

[0030] The compensation parameters corresponding to the priority coefficients are determined according to the preset compensation matching relationship, and the comprehensive selection parameters are corrected according to the compensation parameters.

[0031] Optionally, the step of selecting a placement queue for the job to be moved from other fully loaded queues includes:

[0032] The request moving operation that currently needs to select a placement queue is defined as a determination operation, and the single moving capacity of the determination operation is defined as a determination capacity;

[0033] A full-load queue whose remaining capacity is not less than the determined capacity is defined as an available queue, and the remaining capacity after the addition is determined by performing a difference calculation based on the remaining capacity of the available queue and the determined capacity;

[0034] According to the monomer moving order, the request moving operation adjacent to the request moving operation currently required to be selected for placement in the queue is defined as an adjacent operation, and the monomer moving capacity of the adjacent operation is defined as an adjacent capacity;

[0035] Determine whether the remaining capacity after adding is less than the adjacent capacity;

[0036] If the remaining capacity after adding is less than the adjacent capacity, a calculation is performed based on the remaining capacity after adding and a preset first balance coefficient to determine the optimal parameter;

[0037] If the remaining capacity after adding is not less than the adjacent capacity, the excess capacity is determined by performing a difference calculation based on the remaining capacity after adding and the adjacent capacity, and the preferred parameter is determined by performing a calculation based on the excess capacity and a preset second balance coefficient;

[0038] A priority parameter with the largest value is determined according to the sorting rule, and a full-load queue corresponding to the priority parameter is determined as a placement queue.

[0039] Optionally, after the available queues are determined, the data center-based resource scheduling method further includes:

[0040] Determine whether there is an available queue;

[0041] If there is an available queue, the remaining capacity is determined after adding it;

[0042] If there is no available queue, the requested move job that currently needs to select a placement queue is defined as an abnormal job, and it is determined whether the abnormal job is at the front of the single-unit move sequence;

[0043] If the abnormal operation is at the front of the monomer movement sequence, the corresponding combination that meets the requirements will be defined as an invalid combination;

[0044] If the abnormal operation is not at the front of the monomer moving sequence, the required moving operation before the abnormal operation in the monomer moving sequence is defined as the replacement operation;

[0045] According to the monomer movement sequence, the placement queues determined by the replacement operation are canceled from front to back, and the placement queue selection is re-performed, and the available queues are defined again until an invalid combination or an optional combination is determined.

[0046] Optionally, the step of selecting a unique operation combination from the optional combinations includes:

[0047] Counting the required move operations within the selectable combination to determine the required move quantity;

[0048] Counting different determined placement queues to determine the target queue number;

[0049] Calculate and determine the operational feasibility coefficient based on the required movement quantity, the target queue quantity and the preset weight coefficient;

[0050] The operational feasibility coefficient with the largest value is determined according to the sorting rule, and the optional combination corresponding to the operational feasibility coefficient is determined as the operational combination.

[0051] Optionally, after the operational feasibility coefficient is determined, the resource scheduling method based on the data center further includes:

[0052] Determine whether there are at least two optional combinations with the largest and identical operation feasibility coefficients;

[0053] If there are not at least two optional combinations with the largest and identical operation feasibility coefficients, then determine the operation combination based on the optional combinations;

[0054] If there are at least two optional combinations with the largest and identical operation feasibility coefficients, the job generation time of the job required to be moved is obtained, and the end job time of the corresponding job at the end of the placement queue is obtained;

[0055] Calculate the difference between the terminal operation time and the operation generation time to determine the operation delay duration;

[0056] The average delay time is determined by averaging all the delay times of all the operations, and the overall deviation time is determined by calculating the delay time of the operations and the average delay time;

[0057] The duration selection coefficient is determined by calculation based on the average delay duration and the overall deviation duration, and the duration selection coefficient with the largest value is determined based on the sorting rule, and the optional combination corresponding to the duration selection coefficient is determined as the operation combination.

[0058] In a second aspect, the present application provides a resource scheduling system based on a data center, which adopts the following technical solutions:

[0059] A resource scheduling system based on a data center, comprising:

[0060] The acquisition module is used to obtain the job demand capacity of the current newly added job, the queue consumption capacity of each queue, and the remaining capacity of the queue;

[0061] A processing module, connected to the acquisition module and the judgment module, for storing and processing information;

[0062] A judgment module, connected with the acquisition module and the processing module, for judging the information;

[0063] The processing module defines the queues whose remaining capacity is not less than the job demand capacity as accommodating queues, and defines the remaining queues as fully loaded queues;

[0064] The judging module judges whether there is a queue that can be accommodated;

[0065] If the determination module determines that there is an accommodating queue, the processing module selects a unique queuing queue from each accommodating queue and adds the current newly added job to the queuing queue;

[0066] If the judgment module determines that there is no queue that can be accommodated, the processing module defines the queue with the largest remaining capacity as the candidate queue, and performs a difference calculation based on the remaining capacity of the candidate queue and the required capacity of the job to determine the required vacant capacity;

[0067] The processing module defines each job in the candidate queue as a candidate move job, and enables the acquisition module to acquire the single move capacity of each candidate move job;

[0068] The processing module determines a moving operation combination by performing any number of random combinations according to the candidate moving operations, and determines the combined moving capacity by summing up the individual moving capacities of the candidate moving operations in the moving operation combination, and defines a moving operation combination whose combined moving capacity is not less than the required empty capacity as a meeting the requirements combination, and defines the candidate moving operations in the meeting the requirements combination as the required moving operations;

[0069] The processing module sorts the individual moving capacities of the required moving operations from large to small to determine the individual moving order, and selects the placement queue of the required moving operations from the remaining full-load queues in turn according to the individual moving order, and defines the combination that meets the requirements as an optional combination when each required moving operation selects a placement queue;

[0070] The processing module selects a unique operation combination from the optional combinations, and controls each requested move operation to move to a corresponding placement queue according to the operation combination, and adds the current newly added operation to the candidate queue.

[0071] In summary, the present application includes at least one of the following beneficial technical effects:

[0072] When adding new jobs, the remaining capacity of each queue can be analyzed. When each queue cannot accommodate new jobs, the movement of each queue can be analyzed to try to make room for new jobs to enter, thereby reducing the waste of resources.

[0073] In the case where there is a queue that can accommodate new jobs, the specific situation of the queue can be analyzed to determine a more suitable queue for the new job to enter, so as to facilitate subsequent processing of the job;

[0074] When there are multiple moving plans that can meet the requirements, the most appropriate moving plan can be selected according to the specific circumstances of the operation to be moved, so as to facilitate the overall operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] Figure 1 It is a flow chart of a resource scheduling method based on a data center.

[0076] Figure 2 It is a flow chart of the queue determination method.

[0077] Figure 3 It is a flow chart of the comprehensive selection parameter correction method.

[0078] Figure 4 is a flow chart of a placement queue determination method.

[0079] Figure 5 It is a flow chart of the abnormal operation analysis and processing method.

[0080] Figure 6 is a flow chart of the operation combination selection method.

[0081] Figure 7 It is a flow chart of the multi-operation combination screening method.

[0082] Figure 8 It is a module flow chart of a resource scheduling method based on a data center. DETAILED DESCRIPTION

[0083] In order to make the purpose, technical solutions and advantages of this application more clear, the following Figure 1-Figure 8 It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0084] The embodiments of the present application are further described in detail below in conjunction with the drawings in the specification.

[0085] The embodiment of the present application discloses a resource scheduling method based on a data center. When a user adds a new job task, the method analyzes whether each queue can accommodate the task. When all queues cannot accommodate the new job, the method analyzes whether the new job can be placed by moving the jobs in the queue. When a solution that meets the requirements exists, the corresponding job is moved to make room for the new job, thereby reducing the waste of storage resources in the queue.

[0086] Reference Figure 1 The method flow of the resource scheduling method based on the data center includes the following steps:

[0087] Step S100: Obtain the job demand capacity of the current newly added job, the queue consumption capacity of each queue, and the queue remaining capacity.

[0088] The currently newly added job is the job that the user has just submitted and needs to be queued. The job demand capacity is the space capacity required by the job in the queue, the queue consumption capacity is the space capacity occupied by the jobs already in the queue, and the queue remaining capacity is the space capacity that can be occupied by new jobs in the queue.

[0089] Step S101: The queues whose remaining capacity is not less than the job demand capacity are defined as accommodating queues, and the remaining queues are defined as fully loaded queues.

[0090] Define queues that can accommodate and fully loaded queues to distinguish whether the queues can accommodate the current newly added jobs, so as to facilitate further analysis in subsequent steps.

[0091] Step S102: Determine whether there is a queue that can be accommodated.

[0092] The purpose of the judgment is to find out whether there is a queue where the current new job can be directly placed.

[0093] Step S1021: If there are queues that can accommodate, a unique queue is selected from each queue that can accommodate, and the current newly added job is added to the queue.

[0094] When there is an accommodative queue, it means that there is a queue for the current newly added job to enter directly. At this time, a unique queuing queue can be determined for the current newly added job to enter. The queuing queue can be determined by randomly selecting from the accommodative queues, or by the method of steps S200-S205. At this time, the current newly added job is added to the queuing queue to realize subsequent processing of the current newly added job.

[0095] Step S1022: If there is no queue that can be accommodated, the queue with the largest remaining capacity is defined as the candidate queue, and the difference calculation is performed based on the remaining capacity of the candidate queue and the job demand capacity to determine the required vacant capacity.

[0096] When there is no queue that can be accommodated, it means that there is no queue for new jobs to enter directly. Further analysis is needed at this time. Define alternative queues to determine the queue that best meets the requirements. The required shifting capacity is the space capacity that needs to be vacated when the alternative queue can meet the requirements for placing new jobs. It is determined by subtracting the remaining capacity of the queue from the job demand capacity.

[0097] Step S103: define each job in the candidate queue as a candidate move job, and obtain the individual move capacity of each candidate move job.

[0098] Define alternative move jobs to distinguish different jobs. The single move capacity is the space capacity occupied by a single alternative move job in the queue.

[0099] Step S104: Perform any number of random combinations based on the alternative moving operations to determine the moving operation combination, and perform summation calculation based on the individual moving capacities of the alternative moving operations in the moving operation combination to determine the combined moving capacity, and define the moving operation combination whose combined moving capacity is not less than the required moving capacity as a meeting the requirements combination, and define the alternative moving operations in the meeting the requirements combination as required moving operations.

[0100] A move operation combination is a combination of any number of alternative move operations. For example, if there are alternative move operations A, B, and C, there are seven corresponding move operation combinations, namely A, B, C, AB, AC, BC, and ABC; the combined move capacity is the overall space capacity occupied by the alternative move operations in the determined move operation combination, which is obtained by adding the individual move capacities of all the alternative move operations in the combination; defining a combination that meets the requirements to distinguish the move operation combinations that can meet the placement of new jobs after moving, so as to facilitate further analysis later, and at the same time defining the corresponding alternative move operations as required move operations to facilitate subsequent movement control of jobs in the queue.

[0101] Step S105: Sort the monomer moving capacities of the required moving operations from large to small to determine the monomer moving order, and select the placement queue of the required moving operation from the remaining full-load queues in turn according to the monomer moving order, and when each required moving operation selects a placement queue, define the combination that meets the requirements as an optional combination.

[0102] The monomer movement order is the order of jobs obtained by sorting the required movement jobs from large to small according to their respective monomer movement capacities, that is, the order in which the required movement jobs need to be processed and analyzed; the placement queue is the target queue to which the required movement job needs to be moved. The determination method can be to analyze based on the monomer movement capacity and the remaining capacity of the queue of each fully loaded queue. When the required movement job can be placed, it can be determined as a placement queue. At this time, a queue is randomly selected from the queues that meet the requirements as the placement queue. At the same time, the method for determining the placement queue can also be the method of steps S400-step S404, which will not be repeated here; when each required movement job chooses to be placed in the placement queue, it means that the determined required movement job can be moved out of the current alternative queue. At this time, the new job can enter the alternative queue after the required movement job is moved out, so that the remaining capacity space of each queue can be effectively utilized. At this time, the corresponding combination that meets the requirements is defined as an optional combination to achieve the distinction between different combinations that meet the requirements, which is convenient for subsequent analysis.

[0103] Step S106: Select a unique operation combination from the optional combinations, and control each requested move operation to move to a corresponding placement queue according to the operation combination, and add the current newly added operation to the optional queue.

[0104] The operation combination is an optional combination for performing operations. When there are at least two optional combinations, the operation combination can be determined randomly or by the method of steps S600-S603. By determining the operation combination, each job that requires moving can be moved so that the new job can be added to the alternative queue.

[0105] Reference Figure 2 The steps of selecting a unique queue from each accommodating queue include:

[0106] Step S200: Obtain the waiting job capacity of the jobs waiting in each queue.

[0107] The waiting job capacity is the space capacity occupied by a single job in each queue.

[0108] Step S201: determining the waiting job capacity with the smallest value according to a preset sorting rule, and defining the waiting job capacity as a reference job capacity.

[0109] The sorting rule is a method of comparing numerical values ​​set by the staff, such as the bubble method. The sorting rule can be used to determine the waiting job capacity with the smallest value. At this time, it is defined as the benchmark job capacity to determine the minimum capacity job among the jobs that need to be executed in the batch, thereby understanding the specific situation of the batch of jobs.

[0110] Step S202: Calculate the difference between the remaining capacity of each queue that can be accommodated and the job required capacity to determine the capacity after adding.

[0111] The post-addition capacity is the remaining capacity after the queue adds the new job, which is determined by subtracting the job demand capacity from the remaining capacity of the queue.

[0112] Step S203: determining the first selection parameter corresponding to the added capacity according to the preset first selection matching relationship.

[0113] The first selection parameter is a parameter that reflects whether the queue selection is appropriate based on the capacity after adding. When the capacity after adding is larger, it means that there is excess space available for placement in the current queue. At this time, the corresponding first selection parameter is also larger. The first selection matching relationship between the two is determined by the staff through multiple tests in advance.

[0114] Step S204: performing a difference calculation based on the added capacity and the reference operation capacity to determine the operation difference capacity, and determining a second selection parameter corresponding to the operation difference capacity based on a preset second selection matching relationship.

[0115] The job differential capacity is the value obtained by subtracting the benchmark job capacity from the added capacity. The second selection parameter is a parameter determined based on the added capacity that reflects whether the queue selection is appropriate. When the job differential capacity is a negative value, it means that the remaining space is likely to no longer be effectively utilized. At this time, the corresponding first selection parameter is small. When the job differential capacity is a positive value, it means that the remaining space is likely to be utilized. At this time, the corresponding second selection parameter is large. The second selection matching relationship between the two is determined by the staff in advance through multiple tests.

[0116] Step S205: Calculate according to the first selection parameter and the second selection parameter to determine a comprehensive selection parameter, determine the comprehensive selection parameter with the largest value according to the sorting rule, and determine the accommodating queue corresponding to the comprehensive selection parameter as the queuing queue.

[0117] The comprehensive selection parameter is the value obtained by adding the first selection parameter to the second selection parameter. The larger the value is, the more reasonable the selected queue is for adding new jobs. Therefore, the comprehensive selection parameter with the largest value can be determined through the sorting rules, so that the only comprehensive selection parameter can be determined.

[0118] Reference Figure 3 After the comprehensive selection parameters are determined, the resource scheduling method based on the data center also includes:

[0119] Step S300: Determine whether there are at least two queues with consistent comprehensive selection parameters and the largest accommodating queues.

[0120] The purpose of the judgment is to find out whether there are multiple queues that meet the requirements and can be accommodated.

[0121] Step S3001: If there are not at least two queues with the same comprehensive selection parameters and the largest accommodating queue, the queue is determined according to the only comprehensive selection parameter.

[0122] When there are not at least two largest accommodating queues with consistent comprehensive selection parameters, it means that there is only one accommodating queue that meets the requirements, and the queue determination can be carried out normally.

[0123] Step S3002: If there are at least two maximum accommodating queues with consistent comprehensive selection parameters, a consumption ratio is determined based on the queue consumption capacity and the queue remaining capacity of each accommodating queue.

[0124] When there are at least two queues with the same comprehensive selection parameters and the largest accommodating queues, it means that there are multiple accommodating queues that meet the requirements. At this time, further analysis and screening are required; the consumption ratio is the ratio of the jobs in the current queue to the queue space, which is determined by dividing the queue consumption capacity by the sum of the queue consumption capacity and the queue remaining capacity.

[0125] Step S301: Count the number of jobs waiting in each accommodating queue to determine the number of queues waiting.

[0126] The number of jobs waiting in the queue is the number of jobs that already exist in the queue.

[0127] Step S302: Calculate and determine the priority coefficient based on the consumption ratio and the number of queues waiting.

[0128] The priority coefficient reflects whether the queue that can be accommodated can give priority to the newly added queue. The larger the coefficient, the higher the priority level. The calculation formula is: , where θ is the priority coefficient, μ is the consumption ratio, M is the number of queues waiting, and α is a preset fixed coefficient for calculation.

[0129] Step S303: determining compensation parameters corresponding to the priority coefficients according to the preset compensation matching relationship, and modifying the comprehensive selection parameters according to the compensation parameters.

[0130] The compensation parameter is a parameter used to correct the comprehensive selection parameter. Different priority coefficients indicate different priorities of queue selection, and the corresponding compensation parameters are also different. By correcting the comprehensive selection parameters with different compensation parameters, a unique queue can be determined. The compensation matching relationship between the priority coefficient and the compensation parameter is set by the staff according to actual conditions. It is only necessary to ensure that the larger the priority coefficient, the larger the compensation parameter. The correction method of the comprehensive selection parameter is to add the compensation parameter to the original comprehensive selection parameter.

[0131] Reference Figure 4 The steps of selecting a placement queue for a job to be moved from other fully loaded queues include:

[0132] Step S400: defining the request move operation that currently needs to select a placement queue as a determination operation, and defining the single move capacity of the determination operation as a determination capacity.

[0133] Define the decision job to determine the required move job that needs to be placed in the queue selection to facilitate subsequent analysis, and define the decision capacity to facilitate subsequent analysis and calculation.

[0134] Step S401: a fully loaded queue whose remaining capacity is not less than the determined capacity is defined as an available queue, and a difference calculation is performed based on the remaining capacity of the available queue and the determined capacity to determine the remaining capacity after adding.

[0135] When the remaining capacity of a full-loaded queue is not less than the determination capacity, it means that the full-loaded queue can be used for the determination job. At this time, it is defined as an available queue to distinguish different full-loaded queues; the remaining capacity after adding is the space capacity remaining after the determination job is added to the available queue, which is determined by subtracting the determination capacity from the remaining capacity of the queue.

[0136] Step S402: defining the request move operation adjacent to the request move operation currently required to be selected for placement in the queue as an adjacent operation according to the unit move order, and defining the unit move capacity of the adjacent operation as an adjacent capacity.

[0137] Defining connected operations and adjacent capacities facilitates analysis of subsequent steps. Taking the monomer movement sequence XYZ as an example, when the operation is determined to be Y, the corresponding adjacent operation is Z.

[0138] Step S403: Determine whether the remaining capacity after adding is smaller than the adjacent capacity.

[0139] The purpose of the judgment is to find out whether the queue can still add adjacent jobs of the maximum capacity required after the judgment job is added.

[0140] Step S4031: If the remaining capacity after adding is smaller than the adjacent capacity, a calculation is performed based on the remaining capacity after adding and a preset first balance coefficient to determine the optimal parameter.

[0141] When the remaining capacity after adding is less than the adjacent capacity, it means that the queue cannot add adjacent jobs after adding the determined job. At this time, there may be a situation where the subsequent capacity cannot be used; the first balance coefficient is a fixed coefficient set by the staff, and the preferred parameter is a parameter that reflects whether the selection of the placement queue is reasonable, which is determined by multiplying the remaining capacity after adding by the first balance coefficient.

[0142] Step S4032: If the remaining capacity after adding is not less than the adjacent capacity, a difference calculation is performed based on the remaining capacity after adding and the adjacent capacity to determine the excess capacity, and a calculation is performed based on the excess capacity and a preset second balance coefficient to determine the optimal parameter.

[0143] When the remaining capacity after adding is not less than the adjacent capacity, it means that the queue can add at least one subsequent job. At this time, the subsequent space that can be used is large, and further analysis is needed. The excess capacity is the value obtained by subtracting the adjacent capacity from the remaining capacity after adding. The second balance coefficient is a fixed coefficient set by the staff. The preferred parameters in this case can be determined by multiplying the excess capacity by the second balance coefficient. It should be noted that the second balance coefficient should be greater than the first balance coefficient. The corresponding specific value is reasonably set by the staff according to the actual situation.

[0144] Step S404: determining a priority parameter with the largest value according to the sorting rule, and determining the full-load queue corresponding to the priority parameter as the placement queue.

[0145] The priority parameter with the largest value can be determined through the sorting rule, that is, the fully loaded queue corresponding to the priority parameter best meets the current determination operation, so that the placement queue can be determined.

[0146] Reference Figure 5 After the available queues are determined, the data center-based resource scheduling method further includes:

[0147] Step S500: Determine whether there is an available queue.

[0148] The purpose of the determination is to find out whether there is an available queue that can be determined as a placement queue.

[0149] Step S5001: If there is an available queue, the remaining capacity after adding is determined.

[0150] When there is an available queue, it means that there is an available queue corresponding to the determination job, and the queue determination can be performed normally.

[0151] Step S5002: If there is no available queue, the requested move job that currently needs to select a placement queue is defined as an abnormal job, and it is determined whether the abnormal job is at the front of the single-unit move sequence.

[0152] When there is no available queue, it means that there is no queue that meets the requirements in the currently remaining full-load queues. At this time, the required movement job is defined as an abnormal job for identification to facilitate subsequent analysis. The purpose of the judgment is to find out whether the job movement requirements can be met by adjusting the required movement job of the queue that has been determined to be placed in front.

[0153] Step S50021: If the abnormal operation is at the front of the monomer moving sequence, the corresponding combination that meets the requirements is defined as an invalid combination.

[0154] When the abnormal operation is at the front of the single-unit movement sequence, it means that all fully loaded queues cannot meet the requirements. At this time, the combination that meets the requirements cannot be moved and can be defined as an invalid combination.

[0155] Step S50022: If the abnormal operation is not at the front of the monomer moving sequence, the required moving operation before the abnormal operation in the monomer moving sequence is defined as a replacement operation.

[0156] When the abnormal operation is not at the front of the single unit movement sequence, it means that the required movement operation in front can be adjusted. At this time, a replacement operation is defined to facilitate subsequent analysis and processing.

[0157] Step S501: cancel the placement queues determined by the replacement operation in order from front to back according to the monomer movement sequence, reselect the placement queues, and redefine the available queues again until an invalid combination or an optional combination is determined.

[0158] The method for canceling the placement queue is as follows: for example, the order of single-body movement is A, B, C, D, and the abnormal job is C, then the corresponding replacement jobs are A and B. At this time, the placement queue determined by A is canceled to reselect the placement queue from the remaining queues. When A is determined, the placement queue of B is reselected. If C is still an abnormal job, the placement queue of A continues to be replaced until there is no placement queue to choose from A. The invalid combination can be determined based on A. This method can be used to continuously select the placement queue of each job until a combination that fully meets the requirements is determined. When there is no optional combination, it means that the new job cannot be added by the movement method. At this time, the resource scheduling optimization cannot be performed and no additional processing is required.

[0159] Reference Figure 6 , the steps of selecting a unique operation combination from the optional combinations include:

[0160] Step S600: Count the required moving operations in the optional combination to determine the required movement quantity.

[0161] The required movement quantity is the number of required movement operations within the determined optional combination.

[0162] Step S601: Count the different placement queues determined to determine the target queue quantity.

[0163] The target queue number is the number of determined placement queues. When there are two or more identical placement queues determined by the requested move operations, only 1 is recorded when counting the target queue number.

[0164] Step S602: Calculate and determine the operation feasibility coefficient according to the required movement quantity, the target queue quantity and the preset weight coefficient.

[0165] The weight coefficient is a fixed value coefficient set by the staff, and the operational feasibility coefficient is a coefficient that reflects the degree of trouble in the actual operation of the determined optional combination. The larger the coefficient, the simpler the overall operation. The specific calculation formula is: , where ρ is the operational feasibility coefficient, I is the number of required moves, U is the number of target queues, and β is the weight coefficient.

[0166] Step S603: determining the operation feasibility coefficient with the largest value according to the sorting rule, and determining the optional combination corresponding to the operation feasibility coefficient as the operation combination.

[0167] The operational feasibility coefficient with the largest value can be determined through the sorting rule, and the corresponding optional combination is determined as the operational combination to facilitate the movement of the job tasks in the queue.

[0168] Reference Figure 7, after the operational feasibility coefficient is determined, the resource scheduling method based on the data center also includes:

[0169] Step S700: Determine whether there are at least two optional combinations with the largest and identical operational feasibility coefficients.

[0170] The purpose of the judgment is to find out whether there are multiple optional combinations that meet the requirements.

[0171] Step S7001: If there are not at least two optional combinations with the same maximum feasible operation coefficients, determine the operation combination based on the optional combinations.

[0172] When there are not at least two optional combinations with the largest and identical operation feasibility coefficients, it means that there is only one optional combination that meets the requirements, and the operation combination can be determined normally.

[0173] Step S7002: If there are at least two optional combinations with the largest and identical operation feasibility coefficients, the job generation time of the job required to be moved is obtained, and the end job time of the corresponding job at the end of the placement queue is obtained.

[0174] When there are at least two optional combinations with the largest and identical feasibility coefficients of the operations, it means that there are multiple optional combinations that meet the requirements. At this time, such optional combinations need to be further screened; the job generation time is the time point when the user adds the request to move the job, and the end job time is the time point when the user adds the job cup at the end of the placement queue.

[0175] Step S701: Calculate the difference between the terminal operation time and the operation generation time to determine the operation delay duration.

[0176] The job delay duration is the time interval between the terminal job time and the job generation time. When the job generation time is after the terminal job time, the corresponding job delay duration is a negative value.

[0177] Step S702: Calculate the average of all job delay times to determine the average delay time, and calculate the overall deviation time based on the job delay time and the average delay time.

[0178] The average delay time is the average of all the determined job delay times. The calculation formula for the overall deviation time is: , where S is the overall deviation duration, S i The time wasted for each task, S Y The average delay time.

[0179] Step S703: Calculate the duration selection coefficient based on the average delay duration and the overall deviation duration, and determine the duration selection coefficient with the largest value based on the sorting rule, and determine the optional combination corresponding to the duration selection coefficient as the operation combination.

[0180] The duration selection coefficient is a coefficient for selecting optional combinations based on time delays. The larger the coefficient is, the more reasonable the optional combination is. The duration selection coefficient is obtained by multiplying the average delay duration by the overall deviation duration and taking the inverse. Therefore, the smaller the average delay duration and the overall deviation duration are, the better. The operation combination is determined by the duration selection coefficient to minimize the impact of job transfer after the job transfer.

[0181] Reference Figure 8 Based on the same inventive concept, an embodiment of the present invention provides a resource scheduling system based on a data center, including:

[0182] The acquisition module is used to obtain the job demand capacity of the current newly added job, the queue consumption capacity of each queue, and the remaining capacity of the queue;

[0183] A processing module, connected to the acquisition module and the judgment module, for storing and processing information;

[0184] A judgment module, connected with the acquisition module and the processing module, for judging the information;

[0185] The processing module defines the queues whose remaining capacity is not less than the job demand capacity as accommodating queues, and defines the remaining queues as fully loaded queues;

[0186] The judging module judges whether there is a queue that can be accommodated;

[0187] If the determination module determines that there is an accommodating queue, the processing module selects a unique queuing queue from each accommodating queue and adds the current newly added job to the queuing queue;

[0188] If the judgment module determines that there is no queue that can be accommodated, the processing module defines the queue with the largest remaining capacity as the candidate queue, and performs a difference calculation based on the remaining capacity of the candidate queue and the required capacity of the job to determine the required vacant capacity;

[0189] The processing module defines each job in the candidate queue as a candidate move job, and enables the acquisition module to acquire the single move capacity of each candidate move job;

[0190] The processing module determines a moving operation combination by performing any number of random combinations according to the candidate moving operations, and determines the combined moving capacity by summing up the individual moving capacities of the candidate moving operations in the moving operation combination, and defines a moving operation combination whose combined moving capacity is not less than the required empty capacity as a meeting the requirements combination, and defines the candidate moving operations in the meeting the requirements combination as the required moving operations;

[0191] The processing module sorts the individual moving capacities of the required moving operations from large to small to determine the individual moving order, and selects the placement queue of the required moving operations from the remaining full-load queues in turn according to the individual moving order, and defines the combination that meets the requirements as an optional combination when each required moving operation selects a placement queue;

[0192] The processing module selects a unique operation combination from the optional combinations, and controls each requested move operation to move to a corresponding placement queue according to the operation combination, and adds the current newly added operation to the optional queue;

[0193] A queuing queue determination module, used for determining a unique queuing queue in the accommodating queue;

[0194] A comprehensive selection parameter correction module can correct the comprehensive selection parameters according to the specific conditions of each queue when there are multiple queues that meet the requirements;

[0195] A placement queue determination module is used to determine the placement queue of each job that requires moving;

[0196] Abnormal job analysis module, which analyzes and handles the situation where there is no available queue;

[0197] An operation combination selection module, used to select a unique operation combination from multiple selection combinations;

[0198] The operation combination screening module is used to screen out a unique operation combination from multiple optional combinations that meet the requirements.

[0199] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional modules is used as an example. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. The specific working process of the system, device and unit described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

Claims

1. A resource scheduling method based on data center, It is characterized in that include: Obtain the job demand capacity of the current newly added job, the queue consumption capacity of each queue, and the remaining capacity of the queue; The queues whose remaining capacity is not less than the job demand capacity are defined as accommodating queues, and the remaining queues are defined as fully loaded queues; Determine whether there is a queue that can be accommodated; If there are queues that can be accommodated, a unique queue is selected from each queue that can be accommodated, and the current newly added job is added to the queue; If there is no queue that can accommodate, the queue with the largest remaining capacity is defined as the candidate queue, and the difference between the remaining capacity of the candidate queue and the job demand capacity is calculated to determine the required vacant capacity; Define each job in the candidate queue as a candidate move job, and obtain the individual move capacity of each candidate move job; Perform any number of random combinations according to the alternative moving operations to determine the moving operation combination, and perform sum calculation according to the individual moving capacities of the alternative moving operations in the moving operation combination to determine the combined moving capacity, and define the moving operation combination whose combined moving capacity is not less than the required moving capacity as the required combination, and define the alternative moving operations in the required combination as the required moving operations; According to the monomer moving capacity of the required moving operation, the monomer moving order is determined by sorting from large to small, and the placement queue of the required moving operation is selected from the remaining full-load queues in turn according to the monomer moving order, and when each required moving operation selects a placement queue, the combination that meets the requirements is defined as an optional combination; Select a unique operation combination from the optional combinations, and control each requested move operation to move to a corresponding placement queue according to the operation combination, and add the current newly added operation to the optional queue; The steps of selecting a unique queue from among the accommodating queues include: Get the waiting job capacity of the jobs waiting in each queue; Determine the waiting operation capacity with the smallest value according to a preset sorting rule, and define the waiting operation capacity as the benchmark operation capacity; The capacity after adding is determined by calculating the difference between the remaining capacity of each queue that can be accommodated and the required capacity of the job; Determine the first selection parameter corresponding to the added capacity according to the preset first selection matching relationship; Perform difference calculation based on the added capacity and the benchmark operation capacity to determine the operation difference capacity, and determine the second selection parameter corresponding to the operation difference capacity based on a preset second selection matching relationship; A calculation is performed according to the first selection parameter and the second selection parameter to determine a comprehensive selection parameter, and the comprehensive selection parameter with the largest value is determined according to the sorting rule, and the accommodating queue corresponding to the comprehensive selection parameter is determined as the queuing queue.

2. The data center-based resource scheduling method according to claim 1, It is characterized in that After the comprehensive selection parameters are determined, the resource scheduling method based on the data center also includes: Determine whether there are at least two queues with consistent comprehensive selection parameters and the largest accommodating queues; If there are no at least two queues with the same comprehensive selection parameters and the largest accommodating queue, the queue is determined according to the only comprehensive selection parameter; If there are at least two queues that have the same comprehensive selection parameters and are the largest accommodating queues, the consumption ratio is determined by calculating the queue consumption capacity and the queue remaining capacity of each accommodating queue; The number of jobs waiting on each accommodating queue is counted to determine the number of queue waiting; The priority factor is determined by calculating the consumption ratio and the number of queues waiting; The compensation parameters corresponding to the priority coefficients are determined according to the preset compensation matching relationship, and the comprehensive selection parameters are corrected according to the compensation parameters.

3. The data center-based resource scheduling method according to claim 1, It is characterized in that The steps of selecting a placement queue for the job to be moved from other fully loaded queues include: The request moving operation that currently needs to select a placement queue is defined as a determination operation, and the single moving capacity of the determination operation is defined as a determination capacity; A full-load queue whose remaining capacity is not less than the determined capacity is defined as an available queue, and the remaining capacity after the addition is determined by performing a difference calculation based on the remaining capacity of the available queue and the determined capacity; According to the monomer moving order, the request moving operation adjacent to the request moving operation currently required to be selected for placement in the queue is defined as an adjacent operation, and the monomer moving capacity of the adjacent operation is defined as an adjacent capacity; Determine whether the remaining capacity after adding is less than the adjacent capacity; If the remaining capacity after adding is less than the adjacent capacity, a calculation is performed based on the remaining capacity after adding and a preset first balance coefficient to determine the optimal parameter; If the remaining capacity after adding is not less than the adjacent capacity, the excess capacity is determined by performing a difference calculation based on the remaining capacity after adding and the adjacent capacity, and the preferred parameter is determined by performing a calculation based on the excess capacity and a preset second balance coefficient; A priority parameter with the largest value is determined according to the sorting rule, and a full-load queue corresponding to the priority parameter is determined as a placement queue.

4. The data center-based resource scheduling method according to claim 3, It is characterized in that After the available queues are determined, the data center-based resource scheduling method also includes: Determine whether there is an available queue; If there is an available queue, the remaining capacity after adding is determined; If there is no available queue, the requested move job that currently needs to select a placement queue is defined as an abnormal job, and it is determined whether the abnormal job is at the front of the single-unit move sequence; If the abnormal operation is at the front of the monomer movement sequence, the corresponding combination that meets the requirements will be defined as an invalid combination; If the abnormal operation is not at the front of the monomer moving sequence, the required moving operation before the abnormal operation in the monomer moving sequence is defined as the replacement operation; According to the monomer movement sequence, the placement queues determined by the replacement operation are canceled from front to back, and the placement queue selection is re-performed, and the available queues are defined again until an invalid combination or an optional combination is determined.

5. The data center-based resource scheduling method according to claim 1, It is characterized in that The steps of selecting a unique operation combination from the available combinations include: Counting the required move operations within the selectable combination to determine the required move quantity; Counting different determined placement queues to determine the target queue number; Calculate and determine the operational feasibility coefficient based on the required movement quantity, the target queue quantity and the preset weight coefficient; The operational feasibility coefficient with the largest value is determined according to the sorting rule, and the optional combination corresponding to the operational feasibility coefficient is determined as the operational combination.

6. The data center-based resource scheduling method according to claim 5, It is characterized in that After the operational feasibility coefficient is determined, the resource scheduling method based on the data center also includes: Determine whether there are at least two optional combinations with the largest and identical operation feasibility coefficients; If there are not at least two optional combinations with the largest and identical operation feasibility coefficients, then determine the operation combination based on the optional combinations; If there are at least two optional combinations with the largest and identical operation feasibility coefficients, the job generation time of the job required to be moved is obtained, and the end job time of the corresponding job at the end of the placement queue is obtained; Calculate the difference between the terminal operation time and the operation generation time to determine the operation delay time; The average delay time is determined by averaging all the delay times of all the operations, and the overall deviation time is determined by calculating the delay time of the operations and the average delay time; The duration selection coefficient is determined by calculation based on the average delay duration and the overall deviation duration, and the duration selection coefficient with the largest value is determined based on the sorting rule, and the optional combination corresponding to the duration selection coefficient is determined as the operation combination.

7. A resource scheduling system based on data center, It is characterized in that include: The acquisition module is used to obtain the job demand capacity of the current newly added job, the queue consumption capacity of each queue, and the remaining capacity of the queue; A processing module, connected to the acquisition module and the judgment module, for storing and processing information; A judgment module, connected with the acquisition module and the processing module, for judging the information; The processing module defines the queues whose remaining capacity is not less than the job demand capacity as accommodating queues, and defines the remaining queues as fully loaded queues; The judging module judges whether there is a queue that can be accommodated; If the determination module determines that there is an accommodating queue, the processing module selects a unique queuing queue from each accommodating queue and adds the current newly added job to the queuing queue; If the judgment module determines that there is no queue that can be accommodated, the processing module defines the queue with the largest remaining capacity as the candidate queue, and performs a difference calculation based on the remaining capacity of the candidate queue and the required capacity of the job to determine the required vacant capacity; The processing module defines each job in the candidate queue as a candidate move job, and enables the acquisition module to acquire the single move capacity of each candidate move job; The processing module determines a moving operation combination by performing any number of random combinations according to the alternative moving operations, and determines the combined moving capacity by summing up the individual moving capacities of the alternative moving operations in the moving operation combination, and defines a moving operation combination whose combined moving capacity is not less than the required empty capacity as a meeting the requirements combination, and defines the alternative moving operations in the meeting the requirements combination as the required moving operations; The processing module sorts the individual moving capacities of the required moving operations from large to small to determine the individual moving order, and selects the placement queue of the required moving operations from the remaining full-load queues in turn according to the individual moving order, and defines the combination that meets the requirements as an optional combination when each required moving operation selects a placement queue; The processing module selects a unique operation combination from the optional combinations, and controls each requested move operation to move to a corresponding placement queue according to the operation combination, and adds the current newly added operation to the optional queue; The steps of selecting a unique queue from among the accommodating queues include: The acquisition module acquires the waiting job capacity of the jobs waiting in each queue; The processing module determines the waiting operation capacity with the smallest value according to a preset sorting rule, and defines the waiting operation capacity as the benchmark operation capacity; The processing module calculates the difference between the remaining capacity of each queue that can be accommodated and the job demand capacity to determine the capacity after adding; The processing module determines the first selection parameter corresponding to the added capacity according to the preset first selection matching relationship; The processing module performs a difference calculation based on the added capacity and the reference operation capacity to determine the operation difference capacity, and determines a second selection parameter corresponding to the operation difference capacity based on a preset second selection matching relationship; The processing module calculates according to the first selection parameter and the second selection parameter to determine the comprehensive selection parameter, determines the comprehensive selection parameter with the largest value according to the sorting rule, and determines the accommodating queue corresponding to the comprehensive selection parameter as the queuing queue.