Joint flow routing-scheduling for energy efficient software defined data center networks: A prototype of energy-aware network management platform

Data centers are a cost-effective infrastructure for hosting Cloud and Grid applications, but they do incur tremendous energy cost and CO2 emissions. Today׳s data center network architectures such as Fat-tree and BCube are over-provisioned to guarantee large network capacity and meet peak performance requirement. Networks suffer from inefficient power usage when data center traffic is not high. A solution to this problem is the adoption of network management platform such as OpenNaaS, which can be augmented with energy-aware capabilities. We developed a component for energy monitoring and routing in OpenNaaS. Energy-aware OpenNaaS can support different types of OpenFlow controller; it inherits and enhances network management capabilities, e.g. dynamically obtaining power and topology.In this paper we also discuss the evaluation and selection of energy-aware routing strategies based on an initial prototype of energy-aware OpenNaaS. The target strategies are fine-grained as they combine flow routing algorithms that make routing decisions for the flows and flow scheduling algorithms that schedule the flows on the same link. Differently from previous routing work which focuses on power-minimization problem in data center networks, we aim to optimize energy consumption. Our simulation shows that the combination of priority-based shortest routing and exclusive flow scheduling achieves about 5%–35% higher energy efficiency without performance degradation.