Energy-efficient cloud infrastructure for IoT big data processing

Abstract

Internet of Things (IoT) along with Big Data Analytics is poised to become the backbone of Smart and Sustainable Systems which bolster economic, environmental and social sustainability. Cloud-based data centers provide computing power to churn out valuable information from voluminous IoT data. Multifarious servers in the data centers turn out to be the black hole of superfluous energy burn contributing to 23% of the global Carbon dioxide (CO2) emissions in ICT industry. IoT energy concerns are addressed by researches carried out on low-power sensors and improved Machine-to-Machine communications. However, cloud-based data centers still face energy–related challenges. Virtual Machine (VM) consolidation is an approach towards energy efficient cloud infrastructure. Although several works show convincing results of the potential of VM consolidation in simulated environments, there is inadequacy in terms of investigations on real, physical cloud infrastructure for big data workloads. This work intends to evaluate dynamic VM consolidation approaches by combining algorithms from literature. An open source VM consolidation framework, Openstack NEAT is adopted and experiments are conducted on a Multi-node Openstack Cloud with Apache Spark as Big data platform. This work studies the performance based on Service Level Agreement (SLA) metrics and energy usage of compute hosts. The corresponding results are presented based on which the best combination of algorithms is recommended.