Model-driven network management

Despite remarkable developments in open networking and SDN, one of the most crucial elements of operating any network, the management plane, remains relegated to an afterthought. As the control and data planes open up, users are still firmly locked into a myriad of proprietary CLIs, APIs, and extensions to configure and monitor the network.

In this talk, we will describe a new way of managing, monitoring, and testing networking systems based on different types of data models. We describe our approach at Google to use network models to develop intent-based management systems that fundamentally change the way configuration management is done, while also enabling greater velocity and safety in deploying changes. We also discuss our experiences with developing device-level data models to enable vendor-independent management interfaces that allow operators to more easily build automated management systems and validate correct behavior of the network.
Anees Shaikh is with the Network Infrastructure team at Google where he works on management automation, cloud networking, and network reliability in Google's production networks. Prior to joining Google, he was the Chief SDN Architect at IBM where he was responsible for IBM's software-defined networking product architecture and technical strategy, including leading IBM's open source networking engagements. For over a decade, Anees was with the IBM T.J. Watson Research Center, publishing widely in the areas of networking, cloud computing, and system management, and has been an invited speaker in numerous industry forums. Anees holds a Ph.D. in CSE from the University of Michigan.