Mid to large enterprise campus networks present complex operational requirements: The network must operate reliably and provide high-performance connectivity while enforcing organizational policy. It must also provide isolation across complex boundaries, yet remain easy to manage. All the while, operational and capital costs must be kept low. Software Defined Networking (SDN) has the potential to provide a principled solution to these complex operational challenges. However, most existing work to leverage SDN (e.g., [2, 4, 5, 9]) has so far assumed a full deployment of SDN switches. Unlike datacenter networks, enterprise network upgrade begins not with a green field, but with the existing deployment and is typically a staged process. Budgets are constrained, and only a part of the network can be upgraded at a time—SDN deployment in the enterprise is no exception. The realities of network upgrade and the operational challenges facing existing networks lead us to question: (i) What are the benefits of upgrading to a partial SDN deployment? (ii) How do the benefits of principled network orchestration depend on the location of SDN switches? (iii) Given budget constraints, what subset of legacy switches or routers should be SDN upgraded to maximize benefits? To answer these, in this talk we present Panopticon, an architecture and methodology for aiding operators in planning and operating networks that combine legacy switches and routers and SDN switches. We call such networks transitional networks. We show how Panopticon exposes an abstraction of a fully-deployed SDN in a partially upgraded transitional network, where the SDN benefits extend potentially over the entire network. Panopticon overcomes many of the limitations of current approaches for transitional SDN deployments, which we now briefly review.

Top