At the NFV World Congress, the talk in the corridors was that the NFV industry was having hard times, and that it has hit a rocky patch. These rocks come from two sources: the silos of operational data and the lack of clear telco business cases. Together, they are creating a turning point in the inevitable transition to a cloud-based telco data center. The question is how much pain will be experienced until the industry gets there?
The general sense at NFV World Congress was that the industry was in hard times -- a featured presentation by Intel even started with an explicit statement that the industry was in hard times, but then went on to essentially say that it always darkest before the dawn. Corridor comments even put the recent AT&T ECOMP announcements in a somewhat dark light. One industry thought leader said, "AT&T is doing a service to the industry by stubbing its toes and in doing so finding the rocks so we don't have to." (See AT&T Stresses Its Broader NFV Vision.)
Does this mean that network functions virtualization (NFV) is dead? Probably not. Moving from appliance-based data centers to cloud-based data centers has the same sense of inevitability that the move to TCP/IP had a generation ago. Does it mean that everything is hunky dory and if we just muddle along the way we are, everything will be alright? Also, probably not. So what are the rocks that are in the way? Here are my thoughts on the subject after attending NFV World Congress and participating in a number of conversations.
Some event attendees thought that NFV implementation is more difficult and more complicated than people expected. Where the complication is coming from was exemplified by a lunch-table conversation between leaders of two NFV open source groups. One said that their plan was to do a release, or distribution, by the end of the year without "drivers," and that "eventually we have to decide what drivers to create and then create them so that we can talk to real components." The other person nodded and that led to a period of silence. In other words, they are avoiding the operations data silo problem as long as they can and know that it is a serious problem.
Others talked about NFV just putting all the existing problems into virtualized functions. It seems that there is a growing recognition that operational data silos are a major rock in the road.
Another topic of conversation at the event had to do with the lack of sales and the difficulty of maintaining funding in both large, established vendors and small startups. This led to discussion about the lack of business cases and the fear that exists in telcos surrounding the development of use cases. In a hallway conversation, one person with a large telco pointed to a standards group starting an effort on use cases, but with non-verbals that indicated trepidation.
Another person with a vendor offered an explanation for the trepidation -- fear in telco middle management that if they exposed the extent of the silo problem to senior management, they would fire middle managers they see as responsible for letting it get so bad. At the same time, senior management is concerned about the reaction of the financial markets if the extent of the problem becomes public knowledge.
The good news in all this trepidation is that the rocks are being exposed. Once exposed, a problem can be solved.
Right now, it looks like the telco industry can go in two different directions. It can step up and do what is necessary now to address the silo problem and build business cases that lead to efficient automated networks ready for the challenges of today and tomorrow. Or, it can continue to stay stuck on the rocky road until the pain level gets higher.
From my point of view, the encouraging signs point to the first option -- the recognition of the need to address the silo problem in preparation for/or as the NFV transition is made, and the need to develop business cases.
— Mark Cummings, CTO, Orchestral Networks, special to The New IP