On the Linearity of Performance and Energy at Virtual Machine Consolidation: The CiS2 Index for CPU Workload in Server Saturation
Belén G. Bermejo, Carlos Juiz, Carlos Guerrero · 2018
HPC-based datacenters consume significant energy, despite of a multitude of efforts in reducing it. Virtual Machine Consolidation is an effective technique to minimize the number of physical servers in order to reduce energy consumption, and, consequently, data center costs. However, consolidating virtual machines also reduces performance of servers. Thus, an efficient number of virtual machines consolidated in a physical server is one key issue to optimize the tradeoff between performance and energy efficiency in datacenters. In this paper, we propose a new index, namely CiS2(Consolidation index for Server CPU in Saturation), to quantify the relation of performance degradation and energy efficiency of virtual machine consolidation under parallel workload execution, that is, the tradeoff of performance and energy in consolidation. As a result, CiS2provides an approach to help performance engineers to decide about the suitability of a number of consolidated virtual machines. CiS2could be modified and extended to any other non-functional features for servers.