The key distinction of our research is that it focuses on networks running traditional distributed routing protocols. In an SDN you have significant flexibility over how you decide to route/filter traffic, whereas in a traditional network you have to satisfy intents using the capabilities of existing routing protocols. Obviously SDN's flexibility is what makes it attractive, but there is also (significant) overhead to (partially) switch to this architectures. We recognize that traditional routing protocols still far outstrip SDN in terms of deployment, so there is a real need for intent-driven frameworks that work with traditional routing protocols.
We are not alone in building intent-driven frameworks for traditional routing protocols. Other research (e.g., https://netcomplete.ethz.ch and https://github.com/rabeckett/propane) has also taken an intent-driven approach for traditional routing protocols. What differentiates our work is a focus on _updating_ existing configurations rather than generating new configurations from scratch each time the intents change. Aaron P.S. Thanks for taking the survey! On Fri, Sep 7, 2018 at 11:53 AM Aaron Gould <aar...@gvtc.com> wrote: > Hi Aaron, interesting …making routers do what you intend…hmmm… Sounds > like SDN J …how does what you are doing differ from the > intent-based-controller driven sdn concepts that I hear so much about these > days. > > > > BTW, I did the survey. > > > > - Aaron > > > > *From:* NANOG [mailto:nanog-boun...@nanog.org] *On Behalf Of *Aaron > Gember-Jacobson > *Sent:* Friday, September 7, 2018 8:52 AM > *To:* nanog@nanog.org > *Subject:* Study on configuration change practices > > > > We are a team of networking researchers at the University of > Wisconsin-Madison and Colgate University investigating methods for > automatically synthesizing router configurations from high-level > requirements (or intents). To guide our research, we seek to better > understand the configuration change practices used in production networks. > > We would appreciate if you could take 3 minutes to complete our brief, > anonymous > survey: https://colgate.co1.qualtrics.com/jfe/form/SV_3ee26xayy70jP3D > > To learn more about our research, visit > http://aaron.gember-jacobson.com/research/repair > > Thanks, > Aaron Gember-Jacobson > > *Assistant Professor of Computer Science* > > *Colgate University* >
