Written by Jim Metzler
Jim Metzler is an expert at researching and analyzing emerging technology trends within the networking and service provider space.
The “big bang” approach to NFV allows organizations to develop a platform for rolling out any kind of new virtualization. But it also impedes the ability to adapt to rapidly changing technology. On the other hand, it is easier to develop a business case for a singular, well- defined function using an incremental approach to NFV. The downside is that this approach typically results in more silos within the IT organization.
There are a couple of fundamental approaches to planning NFV. Some companies, like AT&T, are taking a big bang approach, and what that means is they're working to develop a platform that can support any form of virtualization. Other companies, British Telecom, work on a case-by-case basis. There are pros and cons to both approaches. In the case of AT&T and the big bang, some of the pros are once you have this platform, you can very quickly roll out new types of services. Also if you build it correctly, it should have the management, the orchestration security that everything needs. Those are very strong positives.
Some of the negatives of that approach can be extremely difficult to develop a business case for this one platform, that supports any possible kind of virtualization. Also, we're seeing technology change really quickly. We're in the first inning of this game, and it may go to extra innings. So for example of changing technologies, until recently, everybody just assumed that NFV would be focused on virtual machines, on VMs. Over the last six or twelve months, the idea of containers has begun to come to the forefront. So, gee, if I build this platform, based on VMs, how easy will it be to change over to something else if a year or two from now that's a better option? When I think of the incremental approach, it's obviously a lot easier to develop a business case for a one well-defined function. Relatively straightforward.
It's also relatively easy to roll out whatever level of management and orchestration that's acceptable for that. So it's a lot easier to roll that out. On the other hand if you keep doing that, without the integrated platform, you get yet more silos of technology and organizations that's badness. So there's pros and cons to both approaches. One company I've talked to, Telefonica, has kind of stated the ideal. "We're going to go down both paths simultaneously and bring them together when that makes sense." That's clearly a better approach, but it leaves a lot to be determined how you bring these two paths together, sometimes down the road, when that makes sense.