Guarantee-Aware Cost Effective Virtual Machine Placement Algorithm for the Cloud

That the sum of guarantees on bandwidth for any link should be smaller than the link capacity, a prerequisite of offering bandwidth guarantees for tenants in the cloud, is satisfied by virtual machine placement algorithms. The current ones, however, either know nothing about bandwidth guarantees or employ a coarse-grained model for resource abstraction. To solve the problem, this paper first proposes a fine-grained virtual machine placement algorithm which is formulated as a nonlinear program of which the objective is to minimize the number of physical machines used. Specifically, apart from constraints for server resources, we add an additional one for each link to ensure the sum of offered guarantees for each of those links is not greater than its capacity. Further, we devise a heuristic algorithm to address the nonlinear programming problem. Through extensive simulations, we show that our approach is cost-effective, which can reduce the number of physical machines required by 26.17% on average compared with the most recent one.