An Optimized Resource Allcaotion Method In Private Clouds

With the popularity of cloud computing,the efficiency of resource utilization of cloud platforms has attracted much attention in recent years.In traditional framework,the resource allocation problems are broken down into two sub-problems according to their functional processes: virtual machine selection(VMS)and virtual machine placement(VMP).However,such a division can not guarantee the global optimal resource allocation.In this paper,we propose a new resource allocation method for private clouds.Firstly,we formulate the joint virtual machine selection and placement(JVMSP)problem,and prove its NP hardness.Then,we propose the Feedback-Decoupling framework that decouples the JVMSP problem into two sub-problems: Max-Capability and Min-Resource.The Max-Capability problem aims to determine a placement scheme for each PM,making it the largest task processing capability.And the Min-Resource problem aims to choose from the well-placed PMs,minimizing the cost of processing all tasks.We prove that the optimal solutions of the two sub-problems can guarantee the optimal solution of the JVMSP problem.Next,we propose both the optimal-solution algorithms and the heuristic algorithms for each of the two sub-problems.When heuristic algorithms are applied in the framework,we also shows the approximate degree of the solution.Finally,the paper evaluates the proposed framework and algorithms through experimental simulation.The results show that the resource allocation framework and algorithms proposed in this paper significantly improve the resource utilization efficiency of cloud platforms.In addition,they also show better adaptability in different data sets compared with traditional methods.In summary,the main contributions of this paper are as follows:1.This paper formulates the cloud resource allocation problem from a global perspective,and proves that it is an NP complete problem.2.This paper presents a new resource allocation framework and proves that it can guarantee the global optimal resource allocation.For the sub-problems converted from the new framework,this paper presents both optimal-solution algorithms and heuristic algorithms.3.This paper shows the approximate degree of the solution when applying heuristic algorithms in the new framework.4.This paper verifies the new framework and algorithms by simulation which shows they significantly improve the resource efficiency.