The Key Technology Research Of Virtual Machine Resource Scheduling Based On Openstack

Cloud computing is a new IT model. By virtualization technology, cloud computing virtualize the underlying hardware resources which can form a large virtual resource pool, and then providing service to the user by using these resources in a dynamic, free expansion way. Therefore, virtualization technology is the key technology of cloud computing. As more and more people using cloud computing, the data center of the cloud is becoming bigger and bigger. How to improve the allocation of physical resources and scheduling the Virtual Machine (VM) resources by virtualization technology to improve the utilization of the entire cloud data centers, reducing energy costs to achieve green energy resources and environmental protection, which has become an important issue in cloud computing environment.Aiming at the above problems, this thesis applied more extensive research on the widely used OpenStack cloud service build platform technology for data center resources scheduling problem of research. This paper launched a series of following work:(1) In terms of selecting a physical machine using a VM placement strategy based on improved ant colony algorithm for multi-objective. The algorithm is a distributed optimization algorithm, which is conducive to parallel computing and can quickly converge continuously updated by the pheromone to the optimal solution. Experimental results show that this method can balance the conflict between the different goal very well, the power supply in order to make the system produce less resource load and less consumption and ensure better application performance. (2) This thesis proposes a multi-objective optimization approach for VM live migration based on OpenStack. Elaborated the algorithm from four aspects:system resource monitoring, resource scheduling, time of migration, determination of which VM to be choosen and where to place it. On the issue of migration time selected, the system set up a TwoCileDelay way and time series prediction method to determine whether-to migrate VM; on the problem of where to place the VMs, in order to avoid an imbalance of resources, this thesis uses a multi-objective optimization algorithm to select the appropriate physical machine, and in order to avoid bunching effect, this thesis designs of selection algorithm based on the probability. This thesis also designed a VM mail alert program. CloudSim simulation results show that this set of VM resource dynamic scheduling approach can be a good method for real-time scheduling, multi-objective optimization algorithm can optimize the performance of the entire data center.(3) This thesis designs a VM resource real-time scheduling platform which is capable of VM lifecycle management:from creating a VM to dynamically adjusting VM resources to the destruction of the VM. The entire VM resource scheduling process achieves the best performance in the entire server cluster based on multi-objective optimization strategy.