Virtual machine placement algorithms to minimize physical machine count

Network services are often provided through virtualization. To efficiently save capital and operational expenditures, virtual machines (VMs) must be optimally placed on physical machines (PMs) to minimize the number of required PMs. Thus, this paper presents fast heuristic VM placement algorithms for a dynamic demand model. The proposed algorithms are evaluated via computer simulation. Results show that solutions close to the strict optimum are obtainable. Further, results show that there is a tradeoff between the number of PMs and the number of migrations.