Abstract
Data center has been working as a cost-efficient infrastructure to store a large amount of data and host service applications. With the virtualization technology, a resource request submitted to data center can be abstracted as a virtual data center (VDC) request which consist of virtual machines (VMs) connected through virtual switches, routers and links with guaranteed bandwidth. As one of the challenges, VDC embedding/provisioning focuses on mapping VDC components onto physical nodes and links in data center. In this paper we study the problem of provisioning/embedding for evolving/dynamic VDC request across federated data centers, such that the total operation cost is minimized. We use the VM migration to reconfigure evolving/dynamic VDC for reducing the total operation cost, as well to consolidate the VDCs on as few servers as possible for reducing the number of active servers and thus lowering energy consumption. We design an efficient framework and algorithm for solving the studied NP-Hard problem. Finally, we evaluate and compare the performance of our proposed approach through extensive simulation experiments. The simulation results show that the proposed approach performs better in terms of lowering total operation cost and energy consumption than existing solution does.