Research On Virtual Resource Allocation Strategy In Cloud Data Centers

With the advent of the cloud era,cloud computing has changed the use of traditional IT infrastructure and software deployment model.Data center as the main carrier of cloud computing,its efficient and reliable operation is particularly important.Virtualization technology provides an effective method for resource allocation and management of cloud data centers.However,the scale of data centers continues to expand,which brings new problems and challenges to the allocation of virtual resources in data centers.First of all,green energy is already the main theme of the current society,how to reduce the energy consumption of the data center has become a hot research.Second,due to the diversity of user requests,there is a widespread waste of resources in data centers.Server resources are not uniform used and resource fragmentation problem needs to be resolved.In addition,network communication of virtual machines(VMs)is also an important issue to be studied.High Communication costs between VMs will affect the quality of service(QoS),which cannot meet the user's real-time requirements.Therefore,this paper focuses on the virtual resource allocation strategy in cloud data centers.Through the analysis of cloud data center resource utilization,we find that there are a lot of unused resource fragments on servers,namely one resource is exhausted while other resources are still remained.It leads to resources on physical machines(PMs)cannot be fully utilized.This article designs a multi-dimensional VM allocation algorithm based on the analysis of the diversity and dynamic of user requests.First,we build a D-dimensional resource state model based on the resource usage of PMs.On the basis of this model,we design the energy-resource status metric(PAR)to deploy VMs,and propose the multi-dimensional virtual resource allocation algorithm based on energy-aware(MRBEA).The algorithm combines the D-dimensional resource state model and the energy model effectively.It can weigh the impact of energy consumption and resource utilization during the dynamic allocation of VMs,so as to achieve the purpose of reducing energy consumption and improving resource utilization.Simulation experiments show that MRBEA in energy consumption and performance have achieved good results.Aiming at the phenomenon of VM communication delay caused by data center topology,we have studied the network topology of data center,and formalize the network communication problem of VMs in data centers.On this basis,we present an energy-saving allocation strategy based on communication-aware(ESBCA).The algorithm have optimized the communication overhead of data centers from the VM migration and placement phases.During the migration phase,ESBCA prioritizes the migration of VM with large traffic from the overloaded PM to reduce traffic across the network.In the VM placement process,we determine the destination PM in two steps.First,we choose the rack to optimize the cost of communication.Then we select the PM with minimum energy consumption in the rack.The experiment results show that ESBCA effectively reduces the cost of data centers without increasing the energy consumption,and improves the migration efficiency of data centers.