Network functions virtualization (NFV) was a primary topic of last week’s Software Telco Congress, exemplified by the use of #theNFVevent as a conference hashtag. And why not: NFV is what makes a software telco, in a sense. The whole principle is that the hardware running these functions should no longer be so important.

With NFV having steeped in the public consciousness for a year, the economic implications are starting to become clearer. In one keynote, Josh Bottum (pictured above), service creation manager at Cisco, argued that NFV isn’t like coupon-clipping, where carriers will get immediate returns by buying equipment more cheaply.

“This is an evolution, not a tactical 90-day or 180-day project,” he said.

That’s going to take a change in mentality, particularly for the larger carriers. Apple and Amazon didn’t get to be so valued in the market just because they cut costs. They fostered cultures where curiosity and inventiveness are encouraged.

Big carriers could stand to do the same, if they want NFV to live up to its potential. “They are, without question, really, really good at running big networks. I just think they need to take off the chains and start to [spread] this appreciation throughout the culture,” Bottum said.

That idea came up later during a short discussion I moderated with Bottum and Sekhar Banerjee, principal product strategist with OpenWave. The question was about which types of carriers might benefit most from NFV. Smaller carriers need to be nimble and more frugal by definition, so they’d likely gravitate toward the idea.

But NFV is being driven by the large carriers, specifically by factions within the carriers that are determined not to let this idea sink. For some, like AT&T, it’s a CEO-driven agenda, Bottum said. It’s why AT&T started Domain 2.0, in a sense demolishing the domain supplier concept to give smaller vendors a foot in the door, he said.

“There’s too much momentum with that [ETSI] group,” Bottum said. “The big guys are leading.”

NFV and Fast-Fail

That connects back to something I heard Telefónica mention in October, at the SDN & OpenFlow World Congress: that NFV is more about hardware reuse than about cheap hardware. If an application turns out to be a dud, the carrier can kill it off and apply the hardware to something better.

Carriers claim they’re ready for that fast-fail kind of environment. They’ll have to be, because they’re competing with over-the-top players that are accustomed to that pace, Banerjee and Bottum agreed.

One place for carriers to prove their fast-fail mettle would be in virtualized customer premises equipment (CPE) — generic set-top boxes running services that exist as virtual machines back in the cloud. Such a CPE doesn’t have to be changed in order to add new services — meaning new services could be rolled out more quickly, leading to revenues more quickly.

That’s a nice short-term gain. The bigger effect might be that the carrier could move quickly to the next idea, creating a faster treadmill of new services, theoretically.

Interestingly, a virtual CPE would probably cost more than a normal CPE, Bottum said. That’s because costs have been wrung out of CPE gear for generation after generation. Off-the-shelf hardware would actually have to catch up.

But there are chances for new efficiencies if CPE functions turn into virtual machines. “We’ve seen applications use 75 percent less CPU and memory in two releases,” Bottum said.