Network engineering and operations today are highly complex because of vendor heterogeneity, sophisticated intended policies, low-level configuration languages, and rich emergent behaviors. This complexity can leave a wide gap between the engineer’s intent and the network’s actual behavior, and it manifests as network outages, security breaches, and an inability to rapidly evolve the network to changing business needs.
I argue that intent-based networking, which aims to reduce this gap, can be realized by borrowing (and adapting) approaches that have enabled hardware and software engineers to tame their own engineering complexity. I’ll discuss three such approaches: formal validation of behavior, automatic synthesis of low-level constructs from high-level policy specifications, and sophisticated development tools. For each, I’ll sum up the current state of the art, as it applies to networking, and challenges that remain to be solved. Precise and scalable models of the network control plane are a key enabler of all three approaches. Pleasantly, recent work has shown that such models are within reach and readily applicable to real-world networks. I’ll conclude by outlining how these three approaches can be combined to enable highly agile, secure, and reliable networks.
CEO & Co-Founder of Intentionet; Affiliate Professor at University of Washington
Ratul Mahajan is the CEO and co-founder of Intentionet, and an Affiliate Professor at the University of Washington. Prior to that, he was a Principal Researcher at Microsoft Research, where he helped develop several technologies in the areas of software-defined networking, network verification and debugging, and network telemetry that power the world's largest networks. He has been recognized as an ACM Distinguished Scientist, an ACM SIGCOMM Rising Star, and a Microsoft Research Graduate Fellow. His papers have been awarded the ACM SIGCOMM Test of Time award, the IEEE William R. Bennett prize, and several best paper awards at top conferences.