Managing Performance Degradation of Collocated Virtual Machines in Private Cloud
Roozbeh Matloobi, Albert Y. Zomaya · 2016
Virtual machine (VM) consolidation is very popular in cloud computing systems for numerous reasons including improvement of resource utilization, saving of infrastructure costs and ease of management. However, consolidation also causes contention in virtualized environments such as causing to degradation of VM performance. In this paper, we present an effective consolidation system which is able to minimize performance degradation between consolidated VMs. In order to achieve this, we first investigate the relationship between physical resource contention and latency performance degradation for consolidated VMs by comparing VM performance in isolation and virtualization environments. Then, we focus on different existing schemes on the consolidation of VM performance optimization problems in virtualized environments. We attempt experimentally to find a better alternative method to minimize performance degradation and we notice that particle swarm optimization (PSO) has substantial potential. However, PSO must be improved to include redefining the operators and parameters to solve an optimization problem. Experimental evaluations on resource benchmarks using web service stacks, data-centric programming languages, Hadoop and JDK9 indicate that our system has attained encouraging result as compared to previous models.