Research On Virtual Resource Scheduling Strategy For Cloud Computing

With the rapid development of Cloud Computing,research on how to schedule and manage Cloud Computing virtual resources has important meaning to effectively reduce costs,enhance the user experience and ensure the safety and reliability of Cloud Computing services.Hardware resources in data center are virtualized by Cloud Computing as virtual resources in resource pool for unified management and external services.For large-scale Cloud Computing data center,the virtual resources it has to manage and schedule often reach millions level,and at the meanwhile,it also needs to deal with the high intensive Cloud Computing task request,so virtual resource scheduling has become the focus and difficulty in Cloud Computing.At present,the research on virtual resource scheduling under Cloud Computing environment is becoming more and more mature.Many domestic and foreign research's focus are also different,some research's focus are on low cost and quick response,and some researches regard the load balance,service reliability as the goal,and at the same time,there are also studies which focus on green energy.With the expansion of the scale of Cloud Computing,energy consumption and load imbalance problem are becoming more and more prominent,this paper focus on energy saving and load optimization.The virtual resource scheduling problem is a complicated problem,so the virtual resource scheduling problem was divided into two scheduling problems: the virtual machine resource scheduling and the cloud computing task scheduling.In order to realize the goals of energy saving and load optimization,virtual machine scheduling was divided into four stages: Hot host detection,the determining of low load host,the selection of virtual machine which should be migrated,the redistribution of virtual machine to be migrated.In order to measure the energy consumption of the data center and the system load reasonably,this paper established the energy consumption calculation model and load evaluation model.After collected the relevant data of the host,this paper used the partial least squares regression algorithm to detect the hot spots,and used the fixed threshold method and sequential detection method to determine the low load host.After the completion of the above work,this paper established an improved maximum potential migration algorithm for the selection of migrating virtual machines by using the energy consumption calculation model and load evaluation model.After the completion of the selection,this paper established a multi-objective evolutionary algorithm to optimize the energy consumption and load.Finally,conducted the simulation experiments,and then verified the established scheduling strategy.For cloud computing task scheduling,different users have different Qos(Quality of service)requirements,inspired by this,the paper first defined the user's multi-dimensional Qos,then it proposed a genetic ant colony algorithm,and at the same time it integrated the user multi Qos constraints into local and global update,thus it can find the optimal allocation scheme.Finally,the algorithm was validated and compared on CloudSim Platform.The experimental results show that: different scheduling strategy of virtual machines has different advantages.The scheduling strategy(AMOGA)proposed in this paper has better performance than the scheduling strategy of DVFS,LrMmt and ThrMc in terms of energy consumption and load,but the performance in the migration times and average transfer time and SLA violation rate needs to be improved;Compared with simulated annealing algorithm(SA)and basic ant colony algorithm(ACO),the genetic ant colony fusion algorithm(GAACO)proposed in this paper improves the convergence speed and global search ability of cloud computing task scheduling.It also has a better performance in load task scheduling time,cost,quality and service system.Based on the above results,the virtual resource scheduling strategy proposed in this paper can achieve the goal of energy saving and load optimization.