NFV (Network functions virtualization)

NFV Highlight: EE Provides Reality Check

One of the most interesting aspects of the network functions virtualization (NFV) debate is that there are so many (slightly) different views about how virtualization might make an impact on the telecom landscape and influence network planning decisions.

That's why the presentation at the SDN & OpenFlow World Congress by Philip Bridge, senior network architect at EE , sticks in the mind: He gave a very frank view of how he saw virtualization being useful in the short-term that may have left a few NFV cheerleaders a little deflated.

That's not because he doesn't see any value in virtualizing network functions -- he does. But his views on what he is prepared to accept from suppliers, and which functions are ripe for early virtualization, made for every interesting listening.

Bridge noted a number of potential use cases for NFV, including the evolved packet core (EPC), various IMS (IP Multimedia Subsystem) elements, cloud RAN (radio access network) deployments, content delivery networks (CDNs), home gateways, and security devices (such as firewalls).

The mobile packet core has attracted a great deal of attention from NFV proponents, who believe it is part of the network that is ripe for functions virtualization. (See SK Telecom Eyes Virtual Packet Core, Comcast Business Wins Multiple MEF Awards, and Packet Core Looks 'Ripe' for Virtualization.)

But for Bridge, the core of the network is not a place for any experimentation. "That's the part of the network that manages our voice traffic and we have nearly 30 million customers," he noted. If you virtualize the packet core, "you are putting your voice through an embryonic and infantile technology -- that's not going to happen" any time soon, he noted.

Bridge added that it would be "at least a year down the line before we look at this for our network," then noted that there are also investment cycles to be taken into account. "We have just refreshed our network so there is no business case for us" to make any changes to the packet core, he said.

Other areas of the network hold more promise for NFV and software-defined networking (SDN), he noted, including network monitoring, MVNO management, value-added services (which currently involve the deployment of a large number of separate boxes), and security, particularly the introduction of virtual firewalls.

Bridge isn't too bothered at this point about how open any NFV-related technology might be. "We would touch proprietary virtualization," he stated. "I don't care if it's proprietary as long as it reduces complexity -- the challenge is to condense stuff onto fewer boxes."

For the EE man, it's all about figuring out what makes sense. "We're playing with the technology, building sandboxes… It's an evolution, not a revolution," he noted, adding that the combination of NFV, SDN, and Big Data also held "exciting possibilities for extracting and analyzing more detailed intelligence" about the network and applications.

For more on what went down at the SDN & OpenFlow World Congress, see the articles listed below.

Related posts:

— Ray Le Maistre, Editor-in-Chief, Light Reading

DOShea 12/31/2013 | 5:03:30 PM
Network monitoring Network monitoring and traffic management seem like early favorites for NFV activity, and his comments about the core should help cool some of the hype about how revolutionary NFV can be in the short term.
Vishnu Goel 12/27/2013 | 12:22:06 AM
Virtualisation will depend more segregation of hardware and software! I think the way Virtualisation is entering in the field of Computing and Communications, it is the segregation of hardware and software functions is more important than sorting out who goes first for Virtualisation.Commercially the idea is to save on Capex first and achieve efficiency therein.Models like Nicira (acquired by VMWare) may hold the cue for intermediate scenario.Vishnu Goel +919810101238 
Sign In