A Design of Heterogeneous Cloud Infrastructure for Big Data and Cloud Computing Services

Cloud computing and big data are two core services in many organizations. Combining a big data platform, such as Hadoop, into the cloud architecture using virtualization technique will result in losing the performance benefit of MapReduce. Unique for the existing virtualized big data cloud, this work introduces an innovative cloud architecture called the heterogeneous cloud. In the heterogeneous cloud, the big data service is directly deployed onto the bare-metal machines without the help of virtualization. In order to implement such a heterogeneous architecture, we propose a primer architecture design with detailed solutions of potential technical issues and conflicts. With the maximal isolation of computing, network, and storage, the proposed heterogeneous cloud can reach the multifunction of big data and cloud computing with improved performance and energy efficiency.