I am currently studying the use of Optical Switching in the datacenter. At high network demand, the technology allows less costly and more energy efficient datacenter networks compared to traditional electronic (or digital) packet switched networks. The group I am collaborating with developed an optical switch prototype called Mordia, a super fast Optical Circuit Switch for datacenters. I am one of the people responsible for the software infrastructure behind the prototype and my goal is to understand and define the needs of distributed applications from an OCS network. We envision developing software techniques that allow data center applications to seamlessly exploit the scalability and bandwidth offered by optical technologies while maintaing the same (or improving) distributed application performance.
I had the opportunity to work on orchestration and traffic engineering of Wide Area Networks (WAN). Currently, the management of these networks is mostly segregated and inefficient, partly due to the lack of coordination among network devices which forward traffic with "distinct knowledge" of network flows. I.e. routers can identify application-to-application flows as they process IP packets, optical transport network devices (such as Infinera's DTN-X) use SONET/SDH and have a coarse grained view of network flows, whereas ROADMs or MEMS-based optical switches have no knowledge of data streams. As a result, different teams of engineers work at different "layers" (IP, SONET/SDH, Optical WDM) increasing OpEx and preventing efficient and dynamic WANs. We developed a system to solve this problem, and the solution was well accepted by the industry and research community