Three days at the SDN and OpenFlow World Congress outside Frankfurt, and my throat is getting sore. SDN is definitely on the rise, this event is about twice as big as last year’s, or so I’m told by those who were there. With more than 600 delegates and probably 10% of them with a speaking slot, this is a very busy place for exchanging ideas and explaining technology. More so here than the average trade show; this could be due to the fact that the event title is something new and disruptive.
Speaking of that, I noticed that NFV was absent from the event title. An omission which has been corrected in virtually every presentation I attended. Right now NFV is a totally indispensable part of the SDN-OF-NFV trinity. It will be interesting to see how this evolves as NFV starts to wind down at the end of next year. It’s hard to believe NFV was launched less than a year ago.
As usual, Axel Clauberg (Terastream/Deutsche Telekom) was the big star of the event. With the muscles of the big operator and an unprecedented willingness to explain to the world what the important points are in the modern operator’s SDN transformation, people in the industry are eager to listen. And they do. In the questions that followed his keynote, Axel once again revealed the name of the SDN controller being used in his Terastream network —NCS from Tail-f Systems.
With that said, one notion that was loosely repeated in at least four presentations was the comment that “The SDN industry is in an embryonic state, not yet even infantile.” While I understand where this sentiment comes from, I disagree somewhat. We now have two tier one operators who have passed the proof-of-concept stage using NCS as their SDN controller, one with and one without OpenFlow. To call SDN “embryonic” at this point sounds overly modest to me.
There were a few presentations that impressed me including Google’s talk on “Remaking the Network: the Post-OpenFlow Era” and NTT DOCOMO’s “Carriers’ expectation and challenges for Network Virtualization”. Google verified several of the SDN maxims, e.g. the real world utility of the SDN layers (Forwarding Layer, Network Operating System Layer and Specification Layer). They also pointed to some unforeseen bottlenecks, including seemingly simple OpenFlow control applications, there will quickly be a lot of flows and hence a need for big TCAMs.
NTT DOCOMO presented a very real use case with a strong cause. When the big earthquake struck eastern Japan in 2011, only 5% of the 3G voice calls could be completed due to the massive congestion resulting from all the people suddenly trying to call friends and family. In a situation like this, it would make good sense to reallocate a large portion of the call processing capacity from data and multimedia to temporarily boosted voice capacity. NTT DOCOMO showed how they built a prototype call processing service that could increase call processing capacity 5 fold in only 30 minutes. This of course is a real world use case for SDN and NFV in the service of humanity.
In the many conversations I had with NEPs, NETCONF, YANG and transactional management are no longer simply controversial subjects like it was a few years ago. There are still many questions around what they imply and how they differ from other standards and traditions. I think SDN is starting to sink in, and transactional management is understood to be the logical consequence. This is truly encouraging news for the operators and the networking industry as a whole.
UPDATE: Please note that my presentation from the event is now available. You can watch it for free here: “Multi-Vendor SDN for Today’s and Tomorrow’s Networks“