Very recently another European major CSP dramatically scaled back its NFV activities. After a large multi-year effort with both a large internal team and external large vendors, it first moved away from a plan to have one NFV data center for all of Europe to a plan for three data centers. Now, the carrier is closing those three data centers and seeking buyers for them., and, changing leadership of the internal team. It also is postponing the virtualization of mobile voice service for approximately five years, and instead challenging the existing team to make just one NVF implementation work by the end of this year.

While writing this article, news came in about HPE’s shutting down its SDN product line, a 10-percent layoff, not to mention unclear comments about NFV. This development in conjunction with the move to split HPE’s software businesses while merging with MicroFocus is raising uncertainty in some circles.

Other large vendors are showing signs of stress as well. The Ericsson financial situation has been widely reported. Recently well-documented news of a 25-percent staff reduction at Ericsson began to circulate. Not so widely reported has been the similar financial difficulties of the Chinese major vendors and similar staff reductions by them.

Pattern

A highly placed friend in the technical organization of a major vendor wonders if this is just the “slough of despond” that follows the “peak of hype”, and in time all will be well. It is certainly true that NFV has been over hyped. Following the first cycle of Telefonica NFV trouble, notable early stage NFV start-ups failed, and there was uncertainty in the larger vendor community, the feeling at the mid-year 2016 NFV conference in San Jose was worried and depressed. The feeling this year  — just a few months ago — at the same conference was much more optimistic.  Then, the bad news started rolling in.  If it had just been the Telecom Italia pull back, it could have been written off as the slough of despond, but this is something more serious.

Another industry analyst has been very vocal recently saying that the problem is standards. This argument states that if the industry just made the right detailed standards about how all the virtual components interfaced, all would be fine. This highlights the problem.

In the mechanical technology mindset and associated set of habits, one creates detailed standards so that all the gears fit together. Then, you specify in detail a large machine.  You buy the machine; install it; oil it every once in a while; and run it for 50 years.  Software doesn’t work that way.

For conventional software, the large established vendors try to maintain significant proprietary portions of their products. They do this to differentiate their products, value price, and try to lock in their customers. So, for example, 3GPP SA 5 after more than 100 meetings (~4 per year) have at most standardized 40 percent of the North Bound Interface from the element management system (EMS), leaving 60 percent or more proprietary. At the same time, different standards groups rising out of different layers and/or generations of technology create overlapping and competing standards. 

This is not to say that improvements in standards cannot help.  Off and on, over the last couple of years, there have been attempts by the leading telco-focused standards groups to cooperate on trying to create some common data models, or ways of translating between the existing ones. This could be helpful. But fundamentally, standards cannot solve the mechanical mindset and the problems that flow from it.

So the events of the last couple of years show the mechanical mindset and associated habits leading CSPs and their large vendors to try to implement NFV while missing a key ingredient.

Obtaining the Missing Ingredient

The missing ingredient for successful NFV is innovative orchestration software. Innovative software comes from the aformentioned small groups — start-ups and maverick groups in large companies. And end-to-end multi-vendor, multi-domain, orchestration requires fundamental innovation in software architecture.