Virtual Machines Placement for Network Isolation in Clouds

Traffic-aware Virtual Machines (VMs) placement is a key feature for cloud performance isolation. However, variabil-ity hinders a data center to provide network isolation. The existing proposals do not give an answer to a question that how we can obtain resource demand of tenants. To address it, we propose a VMs placement mechanism, which measures resource demand in a server cluster, Incubator before place VMs into a data center. By the stable matching algorithm, we consider resource availability of a data center. We show that our approach decreases cloud oversubscription, and in-creases link utilization through an intensive simulation