Segment routing based energy aware routing for software defined data center

• Published in: Cognitive systems research 2020-12, Vol.64, p.146-163
• Main Authors: Balakiruthiga, B., Deepalakshmi, P., Mohanty, Sachi Nandan, Gupta, Deepak, Pavan Kumar, P., Shankar, K.
• Format: Article
• Language: English
• Subjects: Data Center Network (DCN); Multipath TCP (MPTCP); Open Daylight (ODL); Segment Label Stack (SLS); Segment Routing (SR); Software Defined Data Center (SDDC); Software Defined Network (SDN)
• ISSN: 1389-0417; 1389-0417
• DOI: 10.1016/j.cogsys.2020.08.009

Despite the fact that most of the data centers are software-defined, the multifaceted network architecture and increase in network traffic make data centers to suffer from overhead. Multipath TCP supports multiple paths for a single routing session and ensures proper utilization of bandwidth over all available links. As rise in number of nodes in data center is frequent and drastic, scalability issue limits the performance of many existing techniques. Segment Routing is vibrant in reducing scalability disputes and routing overhead. Segment routing approach combined with MPTCP traffic result in efficient routing approach. The downfall of the link capacity due to drastic incoming traffic remains as a major concern in data center network which enforces preventing link energy depletion due to high network traffic. Our proposed work, segment routing based energy aware routing approach for software defined data center aims to achieve throughput maximization through preserving link residual capacity and proper utilization of links. As well, our approach shows a decrease in length of segment label stack with respect to maximum segment label depth. Analysis is done by comparing the executions of other existing approaches in a single-controller environment with our energy-aware routing approach in a distributed environment. Distributed controller setup prevents network from single point of failure. It helps to prevent controller overhead and provides improved network performance through throughput.