Optimization Algorithm For Task Scheduling In Cloud Computing

With the rapid development of network technology,the generation of cloud computing satisfies the urgent needs for mass data processing capacity.Through the network,services including computing power can be provided to users as a commodity,which makes computing resource available with an on-demand and pay-as-you-go way.Related to the operation efficiency of data centers,task scheduling affects the user's service experience directly.To promote the sustainable development of cloud computing as well as the user's service experience,it is very necessary to develop an efficient and reasonable task scheduling strategy which can truly meet the needs of users.In order to improve the scheduling performance of cloud computing,the researches on independent and related task scheduling algorithm have been made respectively.Besides,aiming at problems of scheduling time and cost which users are most concerned about,improved algorithms are proposed based on common task scheduling algorithms.Firstly,the thesis studies and compares the common task algorithms for independent and related tasks scheduling in cloud computing.And their advantages and disadvantages as well as application characters are analyzed in detail.Secondly,considering scheduling time and cost as well as load balancing of the system,an improved independent task scheduling strategy based on multi-population genetic algorithm is proposed.The algorithm adopts multi-population genetic algorithm instead of genetic algorithm to avoid premature convergence.And in order to improve the search efficiency of the optimal solution,min-min and max-min algorithm are used to initialize the population.For the offspring generated by genetic operation,the Metropolis criteria is used for screening,so that the algorithm can accept a poor solution with a certain probability and avoid falling into the local optimum.Compared with other algorithms,the improved algorithm can reduce scheduling time and cost effectively and can take into account the load balancing of the system.It is an effective task scheduling scheme in the cloud environment and is more suitable for handling quantities of tasks than other algorithms.Finally,in view of the priority constraint between tasks,an improved cost-effective task scheduling algorithm is proposed and the scheduling for related tasks has been converted to large graph processing.In order to explore more possible solutions with higher quality ignored by the deterministic algorithm,this thesis adopts multi-population genetic algorithm to expand the search scope of the optimal solution and designs the fitness function with scheduling time and cost of tasks.In addition,an improved task duplication is proposed to avoid the excessive increase in scheduling cost caused by duplicating tasks blindly.Compared with deterministic scheduling algorithm,the improved algorithm can reduce the scheduling cost effectively with a reasonable scheduling length.