Research On Job Scheduling Algorithm And Implementation Based On Greedy Model In Cloud Environment

Cloud computing is one of the newly-emerging business computing models and thedevelopment of parallel computing, distributed computing and grid computing, whichcan satisfy users’ diverse needs by the way where it make use of sophisticatedvirtualized technology to package resources in data center to provide services throughthe Internet to users. Two key technologies of cloud computing are resource allocationand job scheduling. Its unique business character urges cloud computing providers topay more attention to customers’ satisfaction on the quality of services exceptconcerning the efficiency of resource allocation and job scheduling. Therefore, underthe cloud environment, resource allocation and job scheduling are different from theprevious distributed computing, and have commercial character, service-oriented anduser-centric features.The traditional job scheduling strategy of a cloud environment is inadequate: payattention to efficiency while ignoring the quality of service, pay attention to fairness butefficiency is decreased. This paper systematically studied the cloud computingtechnology, greedy algorithm, and a cloud resource allocation and scheduling problem,a system research of job scheduling algorithm and based on greedy algorithm andgreedy model structure, puts forward the job scheduling strategy based on greedy modelunder cloud environment. The algorithm is distinguished from the traditional jobscheduling algorithm focusing on the efficiency and the recent-proposed oneemphasizing service quality, which establishes dual balance in resource allocation andjob scheduling during cloud environments for the first time, setting balance betweenefficiency and fairness.The first point is that it classifies users in accordance with the QoS preference, bydrawing description, to assign user tasks into different types of queues. Each queue hasa fairness evaluation function to judge the fairness of resource allocation.The second point is that when different types of tasks into the correspondingalgorithm branches, it will allocate resources according to the characteristics of thedifferent queues to achieve the optimal distribution of each branch.Finally, we extend ‘CloudSim’ cloud computing platform, which we add some newmembers and methods of setting and getting these variables to join in the Cloudlet. Inaddition, we overload the method bindCloudletToVM() from the class DatacenterBroker, and recompile the ‘CloudSim’ source code to achieve the proposed job schedulingstrategy based on the greedy algorithm model. With the analysis of results, thealgorithm can achieve double balance, implement user tasks effectively and also providea good quality of service to meet the users’ requirements.