NFV Elements

Verizon Demands Better NFV 'Answers' From Vendors

NICE, France -- TM Forum Live! 2016 -- A senior Verizon executive has accused telecom vendors of failing to come up with the "answers" that operators really need as they introduce software and virtualization technologies into their networks.

Gagan Puranik, Verizon Communications Inc. (NYSE: VZ)'s director of SDN and NFV architecture planning, says most of the virtualization work so far has addressed "low-hanging fruit," such as customer premises equipment and firewalls, and that much of the promises made about the technology have yet to materialize.

"When we talk about decoupling the control and data planes and introducing that kind of philosophy throughout the network, the software from vendors just doesn't meet the scope needed to do this," he told attendees at today's TM Forum Live! event in Nice, France. "That means those planes don't scale independently and we need that."

The executive's remarks came just two weeks after Verizon flagged the completion of an NFV deployment based on OpenStack , an open-source technology, at five of its US data centers. Vendors involved in that project included Big Switch Networks , Dell Technologies (Nasdaq: DELL) and Red Hat Inc. (NYSE: RHT). (See Showdown at the OpenStack Corral.)

One of Puranik's biggest gripes appears to be the lack of progress on the development of microservices, whereby network functions are decomposed into small individual components that operators can reuse and recompose to create customized, scalable applications.

"No one has really developed this on the vendor side," he complained. "They give us big VNFs [virtual network functions] … but we need to develop micro releases so that we can drop incremental upgrades into the network."

Heavy Reading 's Caroline Chappell has previously noted the attractions of microservices, describing the phenomenon as "real cloudification" that would deliver more "extreme business benefits" for telcos. (See The Real NFV Revolution Is 5 Years Away.)

But she has also pointed out that vendors currently have little incentive to develop microservices products given the investments they have made in monolithic network functions.

For more NFV-related coverage and insights, check out our dedicated NFV content channel here on Light Reading.

For Puranik, "micro releases" represent one of the four big promises of SDN and NFV technology, the others being "flexible change management," "automation and analytics" and "shared resources." And while he is relatively happy with the status of the latter, he is asking for greater industry support in other areas.

When it comes to flexible change management, Verizon wants to be able to carry out software upgrades at any time of the day and not be restricted to particular timeslots.

The issue here, however, is to do with people and processes rather than technology, according to Puranik. "The hardest thing is transforming people," he said. "This is a call to action to the TM Forum to help us get there faster and not just to focus on technology but on process and people transformation as well."

The Verizon executive also laments the lack of industry focus so far on automation and analytics, arguing that operators will struggle to develop predictive algorithms -- identifying problems in the making -- without better analytics. "The industry has not focused much on analytics and as a Tier 1 service provider we need that," said Puranik. "Please help us to build artificial brains to create algorithms."

Puranik blames the malaise on a proliferation of standards bodies and open source groups that do not align their activities. "We need alignment so the industry moves faster," he said.

Verizon reinvests about $17 billion in its network each year and is deeply concerned that operational costs are not falling at the pace needed to support growth in demand. "SDN and NFV will help us close the gap but we've not seen that achieve its full potential because of the lack of coordination between standards and open source groups," said Puranik.

— Iain Morris, Circle me on Google+ Follow me on TwitterVisit my LinkedIn profile, News Editor, Light Reading

kq4ym 5/13/2016 | 5:36:33 PM
Re: Better call Gartner! I guess Verizon's got some right to criticize the vendor for going after the "low hanging fruit," although that's probably going to happen in any quickly moving industry expecially where the standards haven't quite been set into mutually agreeable terms and procedural rules quite yet.  It does seem a bit of "cracking the whip" on Verizon's part. 
alangonchar 5/10/2016 | 11:13:42 AM
Better call Gartner! What verizon needs is less survey's (useless except for high level trends) and market research like Garnter and more unbiased, verifiable information based on indepth interviews with the right people. Sorting through a vendors claims versus reality is the only way to go.

They sound surprised that vendors are giving them the acronym based (oh yes we are an end to end solution), marketing stuff!!

Either way I am impressed with Verizons progress and how aggressive they have been in moving forward.
Joe Stanganelli 5/10/2016 | 8:58:39 AM
Re: Whip cracking @brooks: Plus, they're freakin' VERIZON.

If they're that unhappy with what's out there, maybe they out to take some development in house.
Mitch Wagner 5/9/2016 | 11:45:23 PM
Re: Frustration The vision of Docker and other containers is to make software completely platform independent. This has applications to NFV. 
komatineni 5/9/2016 | 8:19:45 PM
Bottomline is does it make business justification. Technology is almost there but the real problem (as mentioned in the article) is around people and processes. this is the most difficult part and will take years to improve. Coming to the technology, the micro services architecture is something which current vendors doesnt see the business case of investing right now. Because this is not a key ask from operators / CSP as the preference is still 3GPP comply or ETSI comply where the 'de-risking' is 'transfer the risk' to vendors in the hundreds of pages of RFP doc. So back to square one. I would think the real innovation in NFV/SDN + Cloud will start to happen when a Web company start to operate at global / regional level in mobile space. Be it for consumer or IOT. 
brooks7 5/9/2016 | 5:28:36 PM
Re: Whip cracking I think there is a fundanmental disconnect here.  These kind of microservices exist.   Just not from vendors that Verizon would recognize nor done in a way that they want to deploy.  Check out Heroku and in particular the Add-ons section: https://elements.heroku.com/addons and


The challenge is that there needs to be architectural planning to be able to coordinate this kind of thing.  Are we talking about REST APIs, Pub/Sub Models with file exchange, Data Base Records, etc etc etc.

That kind of system design needs to come from someone.  Vendors are likely to build some of all of it.  To get a set of components built and deployed in a structured way means that someone will have to define how this will be constructed and connected (how will mini-apps register with each other?).


mendyk 5/9/2016 | 2:09:18 PM
Whip cracking What's somewhat funny here is that it's much easier to get an individual supplier to deliver on what you need rather than expect everything to come from the committee process. There's a reason that this industry continues to revert to variations on the proprietary technology theme -- yes, proprietary solutions can be more expensive and carry the extra baggage of lock-in, etc., but they can happen faster, especially when a big customer brings some pressure to bear.
danielcawrey 5/9/2016 | 1:04:29 PM
Re: Frustration Great insight in this article. We keep hearing about how awesome NFV is going to be, but it doesn't sound like the implementation part is quite solidified yet. I love how Verizon is demanding answers, this sounds like a technical sales call to me. 
Gabriel Brown 5/9/2016 | 11:55:01 AM
Re: Frustration One of challenges to re-writing the VNFs to be more "cloud native" is that the VNF developers don't really know what platform capabilties they are writing to. Things like resiliency and failover, for example -- how does the platform handle that? Or should it be the VNF's responsibility?

This is why it's important for operators to be vocal and clear about what they want, when they want it... and then to actually buy it.

My view is VNFs and platforms will develop in tandem over a few phases. It could be kind of messy for a while, but it's hard to see the market as a whole moving to VNFs or services composed of "micro-services" in short order.  
Mitch Wagner 5/9/2016 | 11:43:41 AM
Frustration If this frustration is shared by many operators, it could derail the NFV movement and send operators back to proprietary solutions. 
Sign In