Uploaded February 2026 | Updated September 2026, 1 week ago
Abstract
In this talk, we explore some of the hidden complexities we observe in running SDN at global-scale, which contribute to both the limited number of SDN deployments and the continued occurrence of outages across network providers. We then present an alternative control plane architecture to address these challenges that we call Decentralized SDN (dSDN). In dSDN, operator-defined SDN control logic runs within routers, eliminating the control plane’s reliance on external infrastructure and restoring fate sharing between control and data planes. We present dSDN as a simpler approach to realizing the benefits of SDN in the WAN. Despite its much simpler design, we show that dSDN is practical from an implementation viewpoint, and can outperform centralized SDN in terms of routing convergence and SLO impact.
Alexander Krentsel:
Alexander Krentsel is a senior researcher in the Systems Research Group at Google, and a fourth-year PhD student at UC Berkeley in the NetSys lab, advised by Scott Shenker and Sylvia Ratnasamy. His research interests are in architecting internet control systems and reasoning rigorously about their correctness and performance. His work at Google focuses on control systems for wide-area networks, and at Berkeley focuses on network-assisted control systems for autonomous vehicles, and agentic systems for infrastructure control system design. Find more information at https://people.eecs.berkeley.edu/~akrentsel/.
nanog.org/events/nanog-96/content/5681
Abstract
In this talk, we explore some of the hidden complexities we observe in running SDN at global-scale, which contribute to both the limited number of SDN deployments and the continued occurrence of outages across network providers. We then present an alternative control plane architecture to address these challenges that we call Decentralized SDN (dSDN). In dSDN, operator-defined SDN control logic runs within routers, eliminating the control plane’s reliance on external infrastructure and restoring fate sharing between control and data planes. We present dSDN as a simpler approach to realizing the benefits of SDN in the WAN. Despite its much simpler design, we show that dSDN is practical from an implementation viewpoint, and can outperform centralized SDN in terms of routing convergence and SLO impact.
Alexander Krentsel:
Alexander Krentsel is a senior researcher in the Systems Research Group at Google, and a fourth-year PhD student at UC Berkeley in the NetSys lab, advised by Scott Shenker and Sylvia Ratnasamy. His research interests are in architecting internet control systems and reasoning rigorously about their correctness and performance. His work at Google focuses on control systems for wide-area networks, and at Berkeley focuses on network-assisted control systems for autonomous vehicles, and agentic systems for infrastructure control system design. Find more information at https://people.eecs.berkeley.edu/~akrentsel/.
nanog.org/events/nanog-96/content/5681
