Network Functions Virtualization (NFV) promises to greatly simplify the management of networks as we know it by leveraging standard x86 server and merchant silicon to turn a broad swath of physical network appliances into applications.
But before any of that can broadly happen the providers of x86 servers and merchant silicon need to find a way to play nice with one another.
With that goal in mind work has begun on a proposed OpenNFV standard that will make it easier to deploy NFV software on any platform regardless of what processor is running inside it.
The development of OpenNFV is being led by Hewlett-Packard. It’s still fairly early in the adoption lifecycle of NFVs. In fact, it may take another good four to five years before NFVs are the standard way to deliver network services.
But the one thing that is for certain is that if NFVs can’t easily move across multiple classes of infrastructure, they may never live up to their full potential.
Nick Ilyadis, vice president and CTO for the Infrastructure & Networking Group at Broadcom, says the OpenNFV initiative is an attempt to make sure that NFVs can run anywhere required. While Broadcom and Intel, for example, are fierce competitors, it’s in the interest of both companies to see the ASICs that tend to dominate networking appliances today replaced by multicore processors in one form or another.
To that end Broadcom recently announced a system-on-a-chip (SOC) that supports OpenNFV. Meanwhile Intel, via its WindRiver subsidiary that provides operating systems for embedded systems, has pledged its support as well.
As much as they might like to dominate the networking space on their own, neither major processor vendor is under the illusion that either one is going to completely dominate the category on their own. But to make sure the networking category shifts in their direction sooner than later they both have a vested interest in working together to see that NFVs adoption occurs.
Of course, there are many networking vendors that would just as soon see NFV adoption come slowly. Most of them are more than willing to promote software-defined network (SDN) as a way to simplify the management of infrastructure. But with the exception of a handful of networking vendors such as Kemp Technologies, NFVs are a still pretty much a theoretical science that for the time being can be ignored.
Nevertheless, providers of IT services have a major interest in the development of NFV software. Not only will NFVs lower the total cost of managing networking; they also provide a mechanism through which new innovative networking services can be more rapidly introduced.
As networking becomes a true service there simply is going to be less of a reason for enterprise IT organization to build their own networks. In fact, SDNs combined with NFVs are projected to be a $20 billion market.
Arguably, compared to the amount of innovations taking place on servers and storage systems the entire networking category has been comparatively stagnant. NFVs represent an innovation that, hopefully, promise to fundamentally reshape the way networks are designed, built and managed. For organizations that make their money providing those network services, this shift to a virtual networking services enabled by NFVs simply can’t happen fast enough.