Research On Resource Load Balancing Scheduling Algorithm Based On Minimum Migration Cost In Cloud Computing Evironment

Cloud computing is an emerging business computing model, it dispatchescomputing user requests to resources pool which is consist of many large-scalecomputers, so that users can obtain the software and storage service by a simpleterminal device. Cloud computing resource scheduling system needs to manage a hugeamount of resources, and must allocate and deploy these heterogeneous resources timelyand effectively to meet the dynamic change of customers’ needs. However, theimbalanced resources distribution will greatly influence the system resource utilization,scalability and user experience. Therefore, how to find a reasonably way scheduling thevirtual machine resources to guarantee the QoS is very important.Different from the traditional scheduling, the basic scheduling unit in the cloudcomputing environment is virtual machine resources, we need to consider the datanetwork transmission delay and a series of influences. But most of the currentscheduling algorithms fail to consider the system variation and historical behavioraldata which causes system load imbalance, can’t meet some specific applicationrequirement.Due to resource scheduling problems involving multi-objective combinatorialoptimization and it proved be a NP Complete problem. Genetic algorithm has goodconvergence effect on optimization problems by using heuristic search. In view of theload balancing problem in VM resources scheduling, this paper presents a schedulingstrategy on load balancing of VM resources based on genetic algorithm. According tohistorical data and current state of the system and through genetic algorithm, thisstrategy computes ahead the influence it will have on the system after the deployment ofthe needed VM resources and then chooses the least-affective solution, through which itachieves the best load balancing and reduces or avoids dynamic migration. Foranalyzing the efficiency of the algorithm, we brings in variation rate to describe the loadvariation of system virtual machines, and it also introduces average load distance tomeasure the overall load balancing effect of the algorithm. The experiment shows thatthis strategy has fairly good global astringency and efficiency, and the algorithm of this paper is, to a great extent, able to solve the problems of load imbalance and highmigration cost after system VM being scheduled. What is more, the average loaddistance does not grow with the increase of VM load variation rate, and the systemscheduling algorithm has quite good resource utility.