Energy-aware Inter-connected Virtual Machine Placement In Cloud Data Center

Virtual Machine Placement(VMP)improves the resource utilization of the Physical Machines(PM)by placing Virtual Machines(VM)appropriately.Place the VM in the PM with low utilization on other PMs,and shut it down to reduce power consumption.For simplicity,the VMP problem was initially modeled under the assumption that VMs are independent of each other.However,in many complicated computing tasks,the computing resources required are far beyond the ability of a single PM.For example,in a scientific workflow,multiple VMs are required to calculate themselves and communicate with each other to complete lots of computing tasks together.However,in existing work,the connection information between VMs during the execution of scientific workflows is rarely studied.Therefore,this paper intends to build a novel connection-aware model for VMP in scientific workflows.Different from existing studies,as the connection information of VMs is considered following the topology of workflows,not only the CPU capacity and memory capacity but also the transmission bandwidth among machines should be considered.The goal of this model is to find the optimal start-up time and running PM for each VM to minimize the total energy consumption over a period of time of the cloud data center.Inspired by the good performance of Ant Colony Optimization(ACO)for solving combinatorial problems,an energy-aware,traffic-aware,connection-aware ant colony optimization(ETCACO)approach is developed for the proposed model.The proposed ETCACO combines ACO with topological solution constructor,greedy placeman,and preprocess strategy.Firstly,ETCACO generates VM orders which satisfy priority constraint based on proposed topological solution constructor.Then,a heuristic method,namely greedy placeman,is designed to map VM orders into VMP schemes.Moreover,ETCACO uses a preprocess strategy to support the processing of multi-batch scientific workflows.Experiments are performed to compare the proposed model with the traditional approach.It is discovered that by considering the connection information,the proposed approach can reduce energy consumption by 7%.