Research On Virtual Machine Deployment And Migration Method For Cloud Computing Center

With the popularity of cloud computing,thousands of servers and virtual machines(VM)exist in large-scale cloud data centers.Currently,cloud centers generally suffer from low utilization of hardware and software resources and high energy consumption.VM placement algorithms can take into account the resource utilization and energy consumption of the entire cloud environment at the time of VM deployment to avoid unreasonable VM deployment schemes.VM migration algorithms can re-integrate resource fragments in the cloud environment to improve resource utilization and reduce energy consumption.Although domestic and international research has achieved some results in solving VM placement and migration problems,they still have some shortcomings.For example,it is difficult to guarantee algorithm stability,operation speed and performance at the same time,and even in highdimensional problems,it is difficult to achieve good performance of all three.Therefore,this thesis proposes a VM placement strategy and a VM migration strategy to solve the problem of high energy consumption and low resource utilization in cloud data centers.The main work of this thesis is as follows:(1)For the placement problem when VMs are created,this thesis proposes a boundary adaptive algorithm ABGSK based on knowledge acquisition and sharing with the optimization goal of minimizing the number of physical servers.ABGSK uses a boundary adaptive strategy(ABS)to achieve dynamic reduction of the search space of the problem.An extended Lagrangian penalty term is also introduced into the objective function to handle the constraints and increase the diversity of individuals in the population.Simulation experiments show that ABS can greatly improve the convergence speed and solution quality of the compared algorithms on small-scale problems,while the improvement is smaller on large-scale problems,which reflects the universality of the adaptive boundary strategy.The ABGSK algorithm achieves excellent results on both small-scale and large-scale problems,and outperforms or equals to other algorithms on small-scale problems,while ABGSK is completely superior to other algorithms on large-scale problems.And the running speed,convergence and robustness of the ABGSK algorithm are significantly better than the other comparative algorithms.(2)For the problem of large amount of resource fragments accumulated during the creation and release of VMs,this thesis proposes a genetic algorithm-based neighborhood search algorithm GNSA to migrate VMs to re-integrate hardware resources in cloud data centers with the optimization objectives of maximizing resource utilization,minimizing energy consumption and minimizing the normalized weighted sum of migration times.Based on the standard genetic algorithm,this thesis designs four neighborhood search operators for the VM migration problem to avoid local optimal solutions,strengthen the search direction and accelerate the convergence speed of the algorithm.Simulation experiments show that GNSA converges quickly on four VM migration problems of different sizes,can enter convergence in no more than 500 iterations,and its adaptation at convergence is optimal among all compared algorithms.In terms of average resource utilization and total power consumption,GNSA significantly outperforms the other comparative algorithms.In terms of number of VM migrations and running speed,GNSA has a higher VM migration rate and takes more time.However,GNSA’s extremely fast convergence speed can compensate for its lack of runtime speed.