Joint study on optimizations of data center deployment, VM assignment and migration

Enterprises build private clouds to provide IT resources for geographically distributed subsidiaries or product divisions. Public cloud providers like Amazon [1] lease their platforms to enterprise users, thus, enterprises can also rent a number of virtual machines (VMs) from their data centers in the service provider networks. Unfortunately, the networks cannot always guarantee stable connectivity for their clients to access the VMs or low-latency transfer among data centers. Usually, latency and bandwidth are in an uncertain network environment. Being affected by background traffics, the network status can be volatile. To reduce the latency of client accesses, enterprises should consider the network status when they deploy data centers or rent virtual data centers from cloud providers. In this paper, we first develop a long-term oblivious data center deployment scheme for an enterprise to meet its client requirements under uncertain network status. Then, we design the optimal VM assignment schemes to assign VMs residing on each data center to each client in the enterprise. To accommodate to the changes of the network status, a VM migration scheme is adopted. The latter two schemes are short-term optimizations given the data center deployment policy. The two-time-scale optimizations work in a joint way, and lay down a framework to help enterprises make better use of private clouds or public clouds.