Hybrid Virtualization Approach for High-Performance Computing Applications in Smart Cities Cloud Infrastructure

Abstract

Nowadays the use of disruptive cloud virtualization technologies oriented towards Smart Cities is not being seized entirely, it is possible to take advantage of new resources management solutions to evolve and turn traditional cloud management processes into a smart quality of service oriented approach. High performance computing (HPC) applications are evolving each day in the cities in order to send more information every moment, HPC applications demand a huge amount of resources. HPC applications have traditionally been executed in Cluster and Grid platforms. Cloud computing has emerged as a platform able to reduce costs while offering on demand access to large amount of computing resources. Hence, it has become attractive to run HPC applications in the cloud. However, the performance penalty imposed by cloud platforms is prohibitive to certain kind of HPC applications. An additional challenge to efficiently run HPC applications in the cloud involves reducing performance interference, which negatively impacts the performance of such applications. In this work, we present a virtual machine placement approach for smart cities infrastructure clouds. We propose the use of a simulated annealing algorithm in order to minimize the interference produced by HPC applications while maintaining the contracted service level agreement (SLA) and minimizing monetary costs. This approach will help Smart Cities cloud resource managers to improve their service and maintain the quality of service demands of the users.