Down, deep in the corners of your data center, underneath the ladder rack where that nasty coil of cable gathers what grime it can from the filtered atmosphere, you pull up a floor tile. There, peering up from the dust bunnies, are the Packet Pushers, analyzers in hand, breathing the pressurized air, and pondering the best way to get traffic from one side of the room to the other. Layer 2 fabric or layer 3? OSPF or BGP? Centralized control plane or distributed? Scale up or scale out? And how much of an oversubscription ratio is acceptable, anyway?
The answer to all of those questions and many more is that, “It depends.” To add to our list of questions today, we ponder another it depends sort of topic.
What is the future of OpenFlow?
Seen as a unifier and enabler in its early days, OpenFlow has come up against some adoption barriers in the form of silicon challenges and vendor-specific extensions that has resulted in a marketplace of OpenFlow options awash in inconsistency. How does OpenFlow rise above this current state of things? Or does it?
What We Discuss
How does the foundation being laid today between the ONF and hardware manufacturers today turn into an OF house being built tomorrow? And what does that house look like? The underlying question is whether or not OF even has a future. There seem to be three different camps.
- No future. Dies when people are tired of playing with the toy.
- Limited future. Useful for corner cases, but never will be useful as a fundamental packet forwarding paradigm.
- The great equalizer. OF will eventually be capable, flexible, and ubiquitous enough to be run complex networks.
We get into these options in nerdy detail — a good discussion.
Links
Show 220 – OpenFlow and Table Type Patterns with Curt Beckmann