Software Defined Survivale Optical Interconnect for Data Centers

Abstract

For service providers, extending the Software Defined Network (SDN) concept from packet switching in Layers 2 and 3 to circuit switching in transport layers is a promising approach to meet high burstiness and high bandwidth requirements. A multi-layer controller that can provide automated controller-based restoration and protection, even for unprotected links in a multi-administrative domain, would be a significant improvement. It would allow service providers to ensure provision of guaranteed Service Level Agreement (SLA) maintenance, with optimal bandwidth usage, high availability and reduced errors. In this thesis, we propose a Software Defined Survivable Optical Interconnect (SDSOI) architecture for Data Centers (DC). This unique architecture will address service providers’ challenges related to bandwidth management, and optimize the time required while interconnecting numerous DCs to meet the high SLA demands. The architecture is built according to the overlay SDN concept, and categorizes the application layers into online, offline and third party applications. The offline application performs the routine DC tasks, while the online application manages various dynamic DC demands. An SDSOI driven Data Center Optical Interconnect (DCOI) can handle the extensive, high quality, on-demand access to the contents. The feasibility of SDSOI is verified and demonstrated using Open Network Operating System (ONOS) as the SDN controller, Mininet as the network emulator and Optical Transport Network (OTN) as the optical framework. Our work primarily focusses on the creation of business applications in the SDSOI model, and the northbound protocols or interfaces used by the applications to interact with the controller. To verify the effectiveness of the proposed SDSOI architecture and its business applications, we simulated ‘Day Night Scheduling’, an application that combines characteristics of offline and online applications. Its primary function is to package and schedule varying DC bandwidths to service providers and optimize bandwidth usage at different times of day, along with the survivability of the interconnects.