An Effective Correlation-Aware VM Placement Scheme for SLA Violation Reduction in Data Centers

Correlation-aware virtual machine placement (CA-VMP) has been demonstrated as one of the most effective energy saving methodologies. Uncorrelated applications are placed together, which based on the off-peak values for application demand, can reduce the number of required servers significantly while meeting the requirements of service level agreement (SLA) at the same time. However, previous works using insufficient constraints for CA-VMP can't provide a good guarantee of SLA performance. In this paper, a detailed analysis is first conducted on the reasons of performance degradation of SLA. To optimize SLA performance, we present a set of placement constraints on the placement capacity of server. We proposed an optimal placement algorithm SSP to achieve a tradeoff between SLA performance and energy cost. According to the simulation experiment results, compared with existing correlation based placement methods, our proposed algorithm provides up to more than five times SLA performance improvement with negligible cost increase.