Research On Virtual Machine Placement Algorithm In Cloud Computing Environment

With the continuous development of the Internet,the demand of massive data for computing is far beyond the computing capacity provided by the traditional computer architecture.The development and promotion of cloud computing not only provides support for this,but also greatly improves resource utilization,reduces energy consumption,and conforms to the concept of national green development.However,data center provides massive resources for a large number of user requests,and the complexity and run time of algorithms are both unimaginable.If multiple virtual machines from the same user request are randomly placed onto different physical machines,the communication between them will inevitably occupy extra bandwidth,which will cause the waste of network resources.When a virtual machine is placed onto the resources from multiple physical machines,the internal communication of virtual machine becomes the communication between hosts,which will make the situation mentioned above worse even lead to network congestion.In addition,the existing academic research is generally based on the premise that the resources required by users will not change,thus occupying resources according to the peak demand in the whole lifetime of user requests,which undoubtedly leads to the waste of resources.This paper studies the problem of finding host for corresponding virtual machine under the location constraints and the problem that user requests occupy resources in their whole lifetime according to peak demand,which will causes the waste of resources.The main contributions are summarized as follows:(1)To solve the problem of virtual machine placement under the location constraints,a placement algorithm based on a compatibility graph is proposed.First of all,the problem is modeled,and the model considering location,computing resources and bandwidth resources is constructed,and the constraints that feasible solution should satisfy are analyzed.Then,the Location-Constrained Virtual Machine Placement(LCVP)algorithm is proposed that can generate corresponding compatibility graph according to the user request.Through this structure,the problem is transformed into a maximum clique problem.By using the structure property of compatibility graph,the problem of finding maximal clique is reduced to finding a maximal clique,which reduces the complexity of the problem.Finally,a heuristic algorithm is used to find a maximal clique with lower resource consumption in the compatibility graph.(2)To solve the problem that user requests always occupy resources in the whole lifetime according to the peak demand,this paper proposes a LC-TD algorithm based on time-varying resources.With the help of time-varying resource demand model,LCVP can be applied to the case of allocating resources in time slots.Each user's demand for resources and the amount of resources that can be provided by the data center are expressed by the number of time slots.Taking advantage of the fact that the time for different user requests to reach the peak demand is often different,multiple user requests can share their own time slots only used in the peak demand,so as to improve the resource utilization.(3)Extensive numerical simulations show that LCVP algorithm and LC-TD algorithm are feasible.Compared with the existing algorithms,they both have shorter run time and higher resource utilization,which can help the data center accommodate more user requests and serve more cloud consumers.