Unlock instant, AI-driven research and patent intelligence for your innovation.

Data center job scheduling method and system based on completion efficiency

A job scheduling and data center technology, applied in the direction of electrical digital data processing, multi-programming device, program control design, etc., can solve the problem that the job completion time cannot be shortened, the job completion time is not considered, and job scheduling affects the computing cluster of the data center Performance and other issues, to achieve the effect of increasing job completion efficiency, improving job completion efficiency, and improving system throughput

Active Publication Date: 2020-10-16
NAT UNIV OF DEFENSE TECH
View PDF5 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

In such data centers, the job scheduling problem of multiple big data analysis jobs is a challenge that affects the performance (such as system throughput) of the computing cluster in the data center
However, this job packaging method does not take into account the completion time of the jobs being packaged together
Packing jobs in this way does not shorten the job completion time

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Data center job scheduling method and system based on completion efficiency
  • Data center job scheduling method and system based on completion efficiency
  • Data center job scheduling method and system based on completion efficiency

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0047] Firstly, the present invention is introduced to inspire the inventive idea of ​​the present invention by analyzing the limitation of the shortest job first strategy and the problem of job packaging.

[0048] Analyze the limitations of the shortest job first strategy, such as figure 2 (a) A case showing the shortest job first policy. In this case, a total of 5 jobs need to be scheduled: the completion time of job 1 is 4 seconds; the completion time of job 2, job 3, job 4, and job 5 is 4.2 seconds. The shortest job first scheduling strategy will schedule job 1 first, and then schedule other jobs. However, job 1 consumes all computing resources of the cluster. This caused other jobs to be delayed. Finally, the average job completion time of the shortest job first scheduling strategy is 7.36 seconds. In this case, the shortest job first scheduling method did not achieve the expected minimum average job completion time. figure 2 (b) shows another heuristic approach to j...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention belongs to the field of big data analysis job scheduling, and discloses a job scheduling method and system based on job completion efficiency. The method comprises the following steps: 1, inputting a batch of jobs and the resource demands and job completion time of the jobs; 2, aligning all the jobs according to the job completion time to generate a plurality of job groups; 3, packaging the jobs in the plurality of job groups to generate a plurality of job sets; 4, scheduling the packaged job sets according to a most dense job set priority scheduling method; and 5, outputting a job scheduling sequence. Through three steps of job completion time alignment, job packaging and scheduling optimization, the job completion efficiency is improved, and the system throughput is improved. The Google cluster data are used for a simulation experiment, and the experiment result verifies the effectiveness of the scheme. Compared with the shortest job priority scheduling strategy, the scheme has the advantages that the average job completion time is reduced by 23.19 percent, and the system throughput is improved by 42.19 percent.

Description

technical field [0001] The invention belongs to the field of big data analysis job scheduling, and in particular relates to a data center job scheduling method and system based on completion efficiency. Background technique [0002] Job scheduling involves how to decide the execution order of jobs: for a batch of jobs, which jobs should be executed in parallel, which jobs should be executed sequentially, and so on. This is a more complex and deeper issue than resource allocation. In large public data centers, it is gradually becoming a trend to execute multiple big data analysis jobs simultaneously. In such data centers, the job scheduling problem of multiple big data analysis jobs is a challenge that affects the performance (such as system throughput) of computing clusters in the data center. [0003] For a single computer, a big data analysis job is a computationally difficult task. This is because the input to the job is a file with a large amount of data. Processing ...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06F9/48G06F9/50G06K9/62
CPCG06F9/4881G06F9/5038G06F2209/484G06F18/23213Y02D10/00
Inventor 胡智尧李东升张一鸣
Owner NAT UNIV OF DEFENSE TECH