CloudEval: A Simulation Environment for Evaluating the Dynamic Cloud VM consolidation

Virtual machine(VM) scheduling algorithms optimize the energy consumption without performance degradation, where cloud providers profited greatly. However, not any algorithms performs the same. To evaluate the power efficiency of such algorithms, various scale of cloud clusters have to be maintained, which is costly and difficult. In this paper, we present a emulation software CloudEval to evalute the performance of migration algorithms of VMs. Comparing to previous cloud simulation environment, we addresses the evaluation of dynamic scheduling of the VMs in the cloud, by a simpler and a non-intrusive manner. Most previous cloud simulation environments encapsulate the simulation process in a close form(eg. the coupling of simulation and evaluation), which is not easy for researchers to design a VM scheduling algorithm and perform the evaluation. Besides, it suggests the clear test cases of workload and energy model, which is also important in VM scheduling algorithm evaluation. Specifically, we address 3 aspects in the evaluation environment: the open architecture of simulation framework for algorithm/strategy design, the evaluation metrics, the simulation cases of workload and mapping of realistic cloud environment. The framework is intended to provide references for algorithm researches.