Cost-efficient Deployment of Big Data Applications in Federated Cloud Systems

Deployment of big data applications requires lot of computing resources as well as bandwidth for data transfer. Federated cloud systems offer a cost-effective solution for such applications requiring storage, processing and computing at scale. However, this comes with the associated energy and bandwidth costs. Our work addresses the problem of minimizing the total operating cost in deploying big data applications in federated cloud systems, including the cost of electricity and bandwidth. We propose an algorithm which considers the communication footprint of the application to minimize the bandwidth cost. The algorithm additionally leverages the diversity in electricity prices across the data centers to minimize the cost of electricity consumption. Experimental results show that the proposed algorithm outperforms other baseline methods in minimizing the operating cost of the cloud provider and also shows the advantages of federation.