Author’s Note: There’s a lot of talk and plenty of noise related to NFV in the telecom industry. It’s a time of market transformation with many executives and marketers making claims for how their companies are doing NFV better than others, or how they are ahead of the competition in one or more ways. I think we are missing an important voice: The CTOs. I am talking with my peers from service providers and suppliers to get a sense of what is real. I am sharing these conversations in this series called The Real CTOs of NFV. The title is fun but the intent is serious. Following is my third Real CTO conversation with Randy Nicklas of Windstream.

Prayson: Randy, I appreciate you taking this time. Please talk a little about Windstream’s role in the NFV transformation in our industry.

Randy: Windstream is currently watchful on NFV, SDN and other developing technologies. We currently do not use NFV techniques unless you stretch the definition to things like Broadsoft feature server platform, which years ago broke out class 5 switch functionality onto, if not Intel servers, then certainly commodity servers, and distributed them across the network.

I don’t know that a lot of people have even said that NFV, in the sense of taking functionality that’s traditionally in embedded software, pulling it out and putting it onto more flexible commodity-like platforms, has actually been part of the telecom industry for a while. I can give other examples as well, certainly other voice or VOIP platforms, the Sonus PSX system for example. Route reflectors running on commodity hardware have been around for a while or at least route servers.

We have an internal team working on both NFV and SDN. And we certainly will see trials in this space in 2016. Windstream’s position on developing technology is to be attentive, send people to conferences, listen to what our competitors are doing, and perhaps most important of all, listen to a variety of vendors, both current and potential vendors, that are developing ideas and concepts in this space.

So, are we a first mover? No, but we’ve definitely been paying attention.

Prayson: Those are good points about how NFV or how virtualization of appliances has been going on for a while. I would add to your list of examples, security. A number of people have been doing hosted security using that approach.

You mentioned that your team is doing some investigation and I’ve been fortunate enough to talk with some of them about that. That leads to my next question in terms of what are the obstacles to NFV transformation that your company is working to address. What areas need to be addressed before Windstream can make that transition to NFV?

Randy: There are a number of obstacles. One that’s common throughout the industry, and not unique to Windstream, is essentially the skill set and organizational structure.

As I’m sure you have experienced with some of your customers, there’s a separation between the IT department and the engineering department with the CTO or the CIO. There’s a boundary of course and a set of hand-offs but not sufficient cross-fertilization.

It’s easy to say we’ll just change our culture. You know, send everybody back to school.

Windstream, like other existing telecom service providers, has a long tail of established ways of doing things, of established systems and platforms on the IT and network side, respectively. Those things aren’t going to go away quickly.

What is beginning to happen is that functionality and product sets are migrating onto both new systems and new network platforms. Let’s use CPE as an example. We have hundreds of thousands of CPE and we’re a $6 billion telecom. If we scale per revenue, smaller companies with revenue measured in hundreds of millions of dollars have proportionately less CPE. AT&T has proportionately more.

There are very large companies like AT&T and others that are making a lot of public announcements in this space. And, it seems they are bending their organizations toward eliminating or at least blurring the boundary between traditional telecom IT and traditional telephone.

No one’s talking about ripping anything out. What we’re talking about is changing how we deliver services. I think it’s appropriate to remind everybody that we’re all going to have this old stuff even as we develop the new stuff. And the balance between the two is going to take years and years to shift. So, when do you think a majority of CPE functionality will be delivered via virtual CPE wherever it’s located? I think we both know it’s going to be measured in years.

There is no big bang. There’s not going to be a big bang at AT&T. And, there’s not going to be the big bang in Windstream or any other company in the industry, unless they are excruciatingly small.

We’ve got all these legacy systems that in some sense need to be adapted because no one’s getting rid of their customer relationship management software or the order entry software, or the billing software for that matter. They all have to be adapted to NFV-capable software platforms and get over to SDN.

Prayson: Those are great points and I’ve heard a lot of people talking about obstacles in terms of organization and skills that you mentioned. Also, the legacy network. Fewer people are talking about that IT aspect and I think that’s very important and critical.

One thing I’ve noticed in my meetings with Windstream is that the people participating are typically much more multi-disciplinary than they were in the past. So, I’m seeing a reflection that people understand that they have to work across the old boundaries. That’s heartening in terms of enabling progress.

Randy: I think there are great examples of people at Windstream who can put on an IT hat when they’ve never actually been paid to do that sort of thing. I don’t see a lot of the diffusion coming from the other side in our organization. And I’ll go back to my previous employer and I didn’t see a lot of it there either. It seems to be the network guys growing more toward the systems guys rather than the reverse. And, I think this is something that we need to change.

Prayson: I believe that in some of our recent meetings there have been some people from IT participating.

Randy: I’m trying to catalyze some of that cross-fertilization at Windstream. What about other companies?

Prayson: I’m starting to see that in some companies. In fact, there’s at least one company that we worked with who has done some reorganization to bring the IT and engineering or CTO groups together under common leadership to enforce that cooperation and break down the silo.

Now, it takes some time for that to propagate down from the higher level. But, that’s a good first step. As you said, AT&T is making those noises but it’ll be interesting to see how a very large organization is able to mechanize that.

I think that it’s generally recognized that to leverage the power of these cloud technologies, like software running on open servers, like ecosystems, like rapid assembly of services, like much more automation – to really embrace these techniques — some of these barriers have to be broken down. That doesn’t make it any easier. But, I am seeing more general recognition that that’s a direction people have to move. It sounds like you’re seeing that also.

Randy: Yes, but it’s not like a freight train about to run us all over.

Prayson: There is difficulty when applying these changes to large companies and that difficulty increases with the size of the company. You mentioned seeing some changes start to roll out, I think you said for Windstream in 2016, so is that when it begins? When does it hit critical mass or become sort of mainstream at least within your company?

Randy: I don’t think it becomes mainstream for quite some time. What does critical mass mean? Does it mean we’re selling a double-digit percentage of our installed service units involved with NFV? Or, does it mean we just got a toe in the water? I think it means more of the latter. So, let’s pick an important service for Windstream and other telecom companies — IP VPNs is a good service to start with.

So, that’s an example of where we might do some basic SDN activities, which for us really means any technique that shortens service delivery intervals as well as subsequent service changes.

We think of NFV in two ways. One is on the infrastructure side and I’ve already alluded to feature servers or policy routing systems being arguably already NFV. You just have to put a sticker on that sort of thing. Another example is session border controllers—it is curious that a lot of my examples come out of the voice world. Call feature servers, policy and routing servers and session border controllers are examples of infrastructure NFV that we already have or we will see in 2016. When we think about actual subscriber-specific NFV — that would probably be on the security side. Virtualized CPE is something we are studying but is further out, and virtual PE routers are something we aren’t convinced are useful to us.

Prayson: As you and the team have started to look at NFV with watchful attentiveness, what is the one thing that’s been most surprising or unexpected?

Randy: I’m a little surprised at the notion that many believe there’s anything really fundamentally new and profound here. I don’t think that there is, but I am ready to be educated. People have been talking about software-based routers on commodity servers, for example, since well before the turn of the century — making a router out of a DEC Alpha workstation is one of many examples. So, there’s a lot of hype around it. Maybe I’m more disappointed than surprised, or maybe I’m just getting old and cynical. I do think some of the element automation techniques pioneered by the hyperscale operators, if not profound, are quite interesting.

I think there have been some interesting fast movers. I’ll use an example, one of your customers — Masergy. I knew a little bit about them back in the day when I worked at my previous employer because they were our customer. They’re an interesting boutique service provider with a small set of services addressing a narrow segment of the market. They had been quite innovative over the years and I think their activities in the NFV space with you guys is just yet another example of that.

Am I surprised that some of this is going as fast as it can? Yes, a little bit. But let’s use CenturyLink. How long do you think it will take them to get to double-digit-number virtual CPEs installed as opposed to the standard installs on the customer prem using Cisco, or Adtran devices or whatever? A long time. But, everything has to start somewhere.

Prayson: To your point about the speed of transition, CenturyLink has been talking publicly for a while, starting out with project Lion and later with other evolutions, about the changes they’re making. But it’s taking them a lot longer than they initially indicated. So, your points about difficulty turning the ship are well taken.

Randy: Every time a new press release comes out or a CEO says something at an earnings release, I hear about it. “Hey, why aren’t we doing this sort of thing?” You have to separate the actual content from the hype. And there’s been a fair amount of hype.

Prayson: There’s been a lot of hype. You mentioned the virtual CPE and that leads to the conversation that people have been having about where to do the virtualization. Should you do it in the classical ETSI NFV manner, which is focused on hosting the virtual functions in a data center or next gen CO? Or, should you push the functions out to the edge of the network and host them on the prem, or some combination thereof?

Randy: I would say it depends. How do you do WAN acceleration in a little cloud pod that’s in the first switch site upstream from the customers’ premises? The answer is you don’t. There are some things that must be done on the far end of the attachment circuit. But there are a lot of things, including the actual CPE layer three functionality, that doesn’t have to be at the far end. So it depends on what it is you’re trying to achieve and the bundle of services you’re trying to deliver to the customer.

I don’t think we’re ever getting to the notion of only some NID-like functionality. Or maybe we are. If it’s just a pure connectivity service then all you need is a measurement point, it doesn’t have to be layer three, you have to be able to essentially loop packet paths and stuff like this at the far end. And that functionality can be quite inexpensive and quite simple.

Let’s be silly and say just the DHCP assignment of addresses into customer LAN at a particular site can come upstream off software running on the virtual machine in a reasonably sized cluster. That’s where we would like to see things happen because otherwise how do you achieve most of the promise of NFV unless you move the actual functionality and the hardware that it’s implemented on upstream?

I know there’s the counter argument. I know you can just run things in software on an Intel processor of some flavor running on the CPE. I think it is desirable to avoid lots of hardware and software functionality residing in the customers’ premises because then you have to take care of it, and this is difficult in part because it is spread over a large area.

So, how is that more like a data center? The thing I like to think about data centers is, yes, they do really impressive things at scale. But, they’re within the confines of, let’s be generous and say it’s diameter of a kilometer, which would be a very large data center. And, they’re within the four walls of a single location, whereas people like us are distributed across hundreds of thousands of square kilometers. To the extent you can reduce the distribution of things that need to be maintained that seems to be an advantage.

Prayson: It makes perfect sense. And that’s very much in alignment with our view that you should put the functions where they need to go based on the requirements of the service.

Randy: I think the preponderance of such functions can be moved upstream.

Prayson: Yes, and as you pointed out, there are some functions that perhaps need to go on at the end of the circuit like the LAN optimization or perhaps security. But, again, that’s based on the requirement to the service. And so I think what you described is a very rational approach.

Randy: I guess the other point is just that there is something here in this NFV concept. The infrastructure will come first and there’s an argument that it’s already here. And it is also the case that we’re not doing anything that would preclude future operations and in the NFV or SDN space, but we are paying attention.

Prayson: As you’re looking at NFV, what’s the one thing that you want NFV to do for your organization?

Randy: That’s easy. What matters most to Windstream is making more money and increasing our gross margin. What is important to us is the cost benefit to the extent that we can reduce operating expense — everybody’s got a CFO. It’s not unimportant to take CapEx out where we can but reducing the on-going cost of delivering our services is incredibly important and that’s a general statement for any business.

Ultimate benefits would include reduction in service delivery intervals, making our networks simpler to diagnose faults, and then having it automatically repaired.

SDN or NFV or anything else are just techniques or tools to those ends. And if they don’t bear fruit in at least one, and ideally multiple areas, then they are not of interest to us.

Prayson: I appreciate your time today, and I hope you have a nice long weekend.

Randy: I will. Take care!

For other entries in the Real CTOs of NFV series, please click here.