Minimizing Communication Cost for Virtual Machine Placement in Cloud Data Center
Kamalesh Karmakar, Rajib Das, Sunirmal Khatua · 2019
In order to achieve high performance and throughput, most of the computing applications nowadays are developed in a distributed computing environment. Such an environment is easily provided by a cloud computing platform. The distributed applications running on cloud demand huge network resources for transfer of data along with computing resources (Virtual Machines). Efficient allocation of virtual machines on hosts can help in the reduction of active hosts which lead to a decrease in deployment cost as well an energy consumption of data centers. We emphasize that while allocating VMs to hosts the communications among the VMs (in the case of distributed applications) cannot be ignored altogether. A placement algorithm which takes into account the cost of physical hosts as well as the communication among VMs can significantly reduce i) the cost of network resource usages ii) energy spent in communication iii) time to complete data transfer and thus improve performance. We have given an ILP formulation of this VM placement problem and then proposed a novel heuristic algorithm to achieve a near optimal solution. We have also analyzed the performance improvement of the proposed algorithm over a few well-known algorithms.