Open adferguson opened 10 years ago
it probably makes sense to work on this in conjunction with a move to NetKAT since, if I remember correctly, NetKAT has a better model for handling topology than NetCore
Almost certainly. We really do have just one "big switch" in the current language.
The good news is that there's no foundational reason why we couldn't add some constraint-based features. That's what LP is for! :-)
On Thu, Jan 2, 2014 at 10:08 PM, Andrew Ferguson notifications@github.comwrote:
it probably makes sense to work on this in conjunction with a move to NetKAT since, if I remember correctly, NetKAT has a better model for handling topology than NetCore
— Reply to this email directly or view it on GitHubhttps://github.com/tnelson/FlowLog/issues/43#issuecomment-31502852 .
how would we even begin to think about circuit provisioning (which is inherently topology-oriented) or WAN traffic engineering in our current system? Flowlog's model is currently very tied to NetCore's "one big switch" abstraction.
for example, ESNet's OSCARS system: