Virtual Machine Placement in Bigdata based Cloud Center

Abstract

With wide application of information systems and mobile devices, vast amount of big data is generated and always managed in a bigdata cloud center. In the cloud center, virtualization is used as a basic resource allocation method and virtual machine placement techniques are applied based on predefined rules. However, some virtual machine placement rules do not consider data replication and I/O requirement of the cloud data center, causing sub-optimal data management outcome. In this paper we propose an optimized virtual machine placement algorithm considering data transfer velocity, cloud storage performance, and network bandwidth. The algorithm is based on heuristics of different tasks data load. Compared with existing virtual machine placement algorithms over extensive experiments indicates that, our proposed solution reduces great energy consumption and needs less migration.