New IP

The Real NFV Revolution Is 5 Years Away

Revolutions do not always have revolutionary outcomes. Monarchs were back on the thrones of England and France not many years after the upheavals that first toppled them. In Russia, the 1905 uprising led to the installation of the Duma, or Russian parliament, but Russians would have to wait another 12 years for the revolution that swept the old regime away.

So it could be with NFV. Skepticism is growing that the first wave of investment in software and virtualization technologies will produce a new telecom order. "If you look at first-generation virtualization, there is not really a lot of automation going on," says Caroline Chappell, practice leader, cloud and NFV, at Heavy Reading . "Operators can still use a lot of their old systems and processes and it's not resulting in the extreme business benefits."

Operators themselves are now playing down expectations surrounding cost. SDN and NFV technology is "not about savings," France's Orange (NYSE: FTE) told Light Reading during a recent conversation. Executives from Deutsche Telekom AG (NYSE: DT) and Verizon Communications Inc. (NYSE: VZ) were similarly downbeat at last week's MPLS/SDN/NFV World Congress in Paris. But other benefits might also be limited, according to Chappell. (See Orange Sours on Cost Benefits of NFV and DT: We Need SDN, NFV to Battle Web Giants.)

"It will give them much more agility than they had, but first-generation virtualization does not on its own deliver OTT [over-the-top] speed to market," she says.

Telcos of the World, Unite!
Executives from the first of two service-provider debates at MPLS/SDN/NFV World Congress, including (from left to right) AT&T's Maria Napierala, Verizon's Chris Emmons, Colt's Mirko Voltolini, Deutsche Telekom's Axel Clauberg, Orange's Francois Bertret, NTT's Kohei Shiomoto and Telecom Italia's Paolo Fasano.
Executives from the first of two service-provider debates at MPLS/SDN/NFV World Congress, including (from left to right) AT&T's Maria Napierala, Verizon's Chris Emmons, Colt's Mirko Voltolini, Deutsche Telekom's Axel Clauberg, Orange's Francois Bertret, NTT's Kohei Shiomoto and Telecom Italia's Paolo Fasano.

For those extreme business benefits, operators may have to go through a second virtualization revolution, introducing what is being called a "microservices" architecture into their networks. In this set-up, a network function would be decomposed into small individual components that an operator could reuse and recompose in many different ways to create customized, scalable applications. "This is real cloudification," says Chappell.

Some of the world's biggest telcos now have microservices in their sights. During a panel debate in Paris, David Hughes, a vice president of engineering at Hong Kong's PCCW Ltd. (NYSE: PCW; Hong Kong: 0008), told attendees that operators would have to move this way to cope with growing customer demand for services that meet very specific requirements. Spain's Telefónica also sees the attractions of using microservices to provide more niche and personalized offerings to its customers.

"When you are dealing with 'head' services it makes sense to have big fat machines," says Diego Lopez, a senior technology expert with Telefónica. "When you are trying to address the long tail you need to go for smaller elements that are faster to deploy and easier to reconfigure."

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

But instigating a microservices revolution might be an even taller order than the initial deployment of SDN and NFV. The uncertainty is whether it will roll on naturally from the first virtualization wave or represent another "discontinuity," as Chappell expects. "First-generation virtualization does not really tackle any of the cloud-native challenges," she says. "At some point there will be another major change."

Service providers are understandably wary. Asked in Paris when he expected to see the first microservices deployments by telcos, PCCW's Hughes cautiously volunteered 2020 as a likely date. Peter Willis, chief researcher of data networks for the UK's BT Group plc (NYSE: BT; London: BTA), reckoned it would be 2021 before they happen. "There is clearly a cost because microservices expose a lot of interfaces externally and create issues around complexity and performance," he says. "But there could certainly be advantages in terms of improving scalability and making it easier to carry out upgrades."

Another obstacle could be vendor unwillingness to support microservices deployments. "There is very little incentive for the vendors to do this because they have got investments in a huge amount of code in existing products," says Chappell.

Indeed, while some operators are putting pressure on vendors to adapt, the development of microservices could have enormous ramifications for the vendor business model. Monolithic network functions priced at a premium would be broken down into what Chappell calls "pinpoint" features carrying much less value. Operators would require only one instance of each feature. "The business model of software is changing," says Chappell. "What you get is much faster innovation, but you also destroy the value in the existing products."

For service providers, one of the biggest challenges of all may be organizational and cultural, rather than technological. Telcos are increasingly talking about having to introduce "DevOps" methodologies into the workplace -- to operate and innovate like fast-moving software companies -- as they redesign their networks. "The whole thing with SDN is making networks look like software, and so we're having to retool the organization to make it look like a software shop," says PCCW's Hughes. (See Telecom Italia Not Ready to Transform, Admits Exec.)

Those DevOps needs may be even more pressing with a microservices architecture, forcing operators to retrain employees, overhaul internal processes and devise new business models. The revolution may be a few years coming, and a messy affair when it arrives, but the cloudification promise makes it an upheaval the industry will not be able to avoid.

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

sj0350 4/3/2016 | 10:51:49 PM
Everything old is...old again NFV may turn into the IMS of this decade: great promise and fanfare, a couple of really good ideas, and fatally bogged down in the tarpits.  One key difference: IMS was killed by vendors who trampled on the seamless interop promised by the standards whereas NFV will be inadvertently strangled by the telcos, who despite all their public pronouncements about refactoring themselves into software companies, are still populated by 40- and 50-somethings who expect to deal with huge vendors over multi-year projects and are punished for breaking things.  In that environment you can forget about agility, low cost and fast time to market.  Yes, NFV lives or dies not by technology but by culture.

As an example: let's say you build an opensource telco widget and a small startup around it.  Google would be able to take that and build glue to do what they needed.  A telco would insist that the vendor do it all, possibly at the latter's cost.  Some startups won't be able to do that.  So now Google succeeds with opensource but the telco ends up buying from BigCorp as the only supplier with deep enough pockets to keep up.  But the project is late and expensive.  A cultural practice ends up having deep commercial impacts.
kq4ym 3/25/2016 | 9:49:29 AM
Re: Step by step It was refreshing to see the other side of the mostly optimistic view of NFV we're used to seeing. And it may well be true that "one of the biggest challenges of all may be organizational and cultural," and we'll have to progress on that front as well as the tech changes coming rapidly.
utae.kim 3/17/2016 | 10:02:05 PM
NFV Benefits for Telco? Really, what do you think about NFV real benefits for Telcos?

It seems to me that the most important thing is to deliver agile serivces(or mircro-services) for end customers and then telcos want to save CAPEX, OPEX with NFV.

At this time, I wanna ask a questions what is agile services(or mircro services). These are a big problem for telcos to solve.

Without solving these, there seem to have many huddles to be innovative with NFV.
Frank Pick 3/17/2016 | 2:47:08 PM
Under a rock? Goodness me people. Are you really living under a rock? The striking irony is an advert on this page to a talk with John Donovan who says "go fast" when it comes to NFV. Yet again our telco industry exhibitis the stodginess it has for years, the one that lost them the field to internet companies, the one that resulted in dwindling revenues, the one that displays again the complete lack of immagination and innovation. I seem to recall somewhere AT&T was heading for 100% NFV in 2020 and closed out last year with 7% of all services virutalized. 

Perhaps it is time for the paradigm shift to take place. MNOs should realize their assets are in their brand and access. Monitze the access by making it available to all and stick to their handset shopfronts, ask yourself a question - what was the last innovative service my network provider actually gave me? For me it was data/internet!

Perhaps it is time to refresh the staff and bring in people that can actually think and deliver or find a managed service provider that can do it for you. We should be ashamed this is the best we can do here. 

Well done John Donovan for displaying the courage to execute, perhaps you could send them a copy of ATT's Domain 2.0 paper for reading. 
vances 3/17/2016 | 1:19:09 AM
Baby Steps The important point about the concept of microservices is that there are many of them.  The collection of these services becomes a sort of cloud operating system.  Today we are doing little more than porting our appliance applications to virtual machines.  In the future we'll write our applications to leverage a wealth of available services, allowing developers to work at a higher level of abstraction.  Examples include replicated data (CRDT), process registries and inversion of control.  We're just getting started.
Juice264 3/14/2016 | 7:04:42 PM
1G, 2G, 3G? I like the introduction of 'first generation' by the panel and writer. Are we headed for 1G, 2G, 3G etc. now for virtualisation like we did with mobile technologies from the late 80's?
msilbey 3/14/2016 | 2:01:50 PM
Step by step NFV has felt like an early step in getting to SDN, and it sounds like microservices may almost be the same. Automation may be the holy grail, but there's a long way to go to get everything in the network to that end goal.
Sign In