Research On Task Scheduling Algorithm Based On Workflow In Hybrid Cloud Environment

Cloud computing,as a technology derived from distributed computing,parallel computing and grid computing,is of great significance to more and more complex scientific computing.As a new type of cloud computing,hybrid cloud is composed of public cloud and private cloud.Scientific workflow is a common organization model of large-scale scientific computing in cloud environment,although its scheduling problem has been studied in many other distributed computing systems,cloud platforms,especially hybrid cloud environment and other traditional distributed computing platforms,still have many different characteristics,so scheduling strategies that perform well on other platforms are not fully applicable to hybrid cloud environment.Therefore,it is of great significance to study workflow-based task scheduling in hybrid cloud environment.This paper focuses on the workflow scheduling problem in hybrid cloud environment and two main factors(time and cost)that affect the quality of service.The main researches and contributions of this paper are as follows:1)The characteristics of hybrid cloud environment is analyzed and the workflow scheduling problem in hybrid cloud environment is modeled.A heterogeneous resource model in hybrid cloud environment is constructed,and the related computing model and billing model of the problem to be studied are given according to the workflow model adopted.2)A scheduling algorithm DCOH(Deadline-constrained Cost Optimization for Hybridcloud)with optimizing cost under deadline constraint is proposed.According to the idea of evolutionary algorithm,a new population initialization scheme,crossover operator,mutation operator and fitness function are designed by using three-segment chromosome coding.Based on a large number of real workflow application simulation experiments,the effectiveness of the proposed algorithm in hybrid cloud environment is proved.3)A multi-objective optimization for Hybrid-cloud(MOH)scheduling algorithm based on evolutionary algorithm is proposed.Based on the idea of DCOH algorithm,a new fitness evaluation function,Pareto technology and crowding distance are introduced for multiobjective scheduling in hybrid cloud environment.Experiments show that the proposed MOH algorithm has better Pareto front than SPEA2* algorithm in many kinds of workflows.