Scheduling method for calculating element

Abstract

The invention relates to a scheduling method for a calculating element. The scheduling method includes the steps that S1, physical machine resources are input; S2, resources of the calculating element required by a user are input; S3, an optimal object is formulation description through a data model; S4, a calculating element deployment scheme is obtained with the grouping genetic algorithm, power consumption, and the resource waste condition and the SLA violating rate are minimum in whole accordingly; S5, the calculating element deployment scheme corresponding to the optimal solution is output. According to the scheduling method for the calculating element, the problem that in the multi-objective optimization process, the influence degrees of multiple optimal objects are uncertain is solved with the improved grouping genetic algorithm, and compared with single object optimization, the scheduling method has the better practicability; the convergence rate of scheduling calculating is obviously increased, and the running quality is guaranteed while energy consumption of a server cluster is greatly saved; the balance performance and the calculating efficiency of server resources are remarkably improved.

Description

Technical field [0001] The invention relates to a computing unit scheduling technology under a cloud platform and an artificial intelligence scheduling method, in particular to a scheduling method of a computing unit that can optimize multiple goals at the same time. Background technique [0002] Computing unit scheduling in distributed computing refers to mapping a group of computing units requested by users to physical machines (ie servers or physical nodes) according to a certain scheduling algorithm, and at the same time, it must meet necessary constraints. The electricity consumption, resource utilization, user experience, and cloud provider revenue brought about by the different mapping between cloud data center computing units and physical machines. Therefore, it is very important to design an effective scheduling algorithm. As the scale of data centers continues to expand, the power consumed by them is growing rapidly. More and more studies have focused on energy saving ...

AI Technical Summary

Problems solved by technology

[0003] At present, the genetic algorithm has at least the following shortcomings in solving the computing unit scheduling problem: 1) The performance of the algorithm depends on the selection of parameters, and improper selection of parameters will seriously affect the quality of the solution

2) The crossover probability and mutation probability are fixed in the process of population evolution, which seriously affects the convergence of the population and is easy to cause premature maturity and fail to reach the global optimal solution

3) The crossover operation and mutation operation in the algorithm randomly select the genes on the chromosome, which is blind and affects the convergence speed of the algorithm

Images