An energy-efficient virtual machine placement and route scheduling scheme in data center networks

• Published in: Future generation computer systems 2017-12, Vol.77, p.1-11
• Main Authors: Yang, Ting, Pen, Haibo, Li, Wei, Zomaya, Albert Y.
• Format: Article
• Language: English
• Subjects: Cloud computing; Data center networks; Energy efficiency; QAP; Routing; Virtual machine placement
• ISSN: 0167-739X; 1872-7115
• DOI: 10.1016/j.future.2017.05.047

The increasing requirements of big data analytics and complex scientific computing impose significant burdens on cloud data centers. As a result, not only the computation but also the communication expenses in data centers are greatly increased. Previous work on green computing in data centers mainly focused on the energy consumption of the servers rather than the communication. However, for those emerging applications with big data-flows transmission, more energy consumption could be consumed by communication links, switching and aggregation elements. To this end, based on data-flows’ transmission characteristics, we proposes a novel Job-Aware Virtual Machine Placement and Route Scheduling (JAVPRS) scheme to reduce the energy consumption of data center networks (DCN) while still meeting as many network QoS (Quality of Service) requirements as possible. Our proposed scheme focuses on not just migrating large data flows, but also integrating small data flows to improve the utilization rate of the communication links. With more idle switches turned off, DCN’s energy consumption will thus be reduced. Besides the data flows’ migration and integration, the Traffic Engineering (TE) technique is also applied to decrease the transmission delay and increase the network throughput. To evaluate the performance of our proposed scheme, a number of simulation studies are performed. Compared to the selected benchmarks, the simulation results showed that JAVPRS can achieve 22.28%–35.72% energy saving while reducing communication delay by 5.8%–6.8% and improving network throughput by 13.3%. •We present a joint optimization energy-efficient approach Job-Aware VM Placement and Route Scheduling (JAVPRS) for data center networks.•Unlike previous approaches, we do not address the issue as a traditional QAP problem. Instead, we fully utilize VM migration technology and communication data flow consolidating technology to minimize the energy consumption and balance the communication load within the data center network.•JAVPRS produces better VM placement and data-flows routing based on the characteristic data flows and the latest status of the network devices.