AT&T's Gilbert Talks Virtualization

During a career that has spanned more than 20 years at A&T, Mazin Gilbert has seen numerous new waves of technology, but none of them is bigger than the telco's current virtualization effort.

This past October, Gilbert was promoted to the position of vice president of advanced technologies at AT&T Labs. In this role, he oversees several key areas of innovation for AT&T including artificial intelligence as well as networking and data analytics. (See AT&T's Gilbert: AI Is Very Real.)

In Part 1 of this Telco Transformation Q&A with Gilbert, which was conducted late last year, he talks about AT&T's virtualization blueprint, its ECOMP (Enhanced Control, Orchestration, Management & Policy) architecture, and his thoughts on what it takes to be a good leader. AT&T Inc. (NYSE: T) is among the worldwide leaders for aggressively adopting SDN and NFV to transform its network and has previously announced a goal of 75% virtualzation by 2020.

Telco Transformation: So what's on AT&T's virtualization roadmap over the next few years?

Mazin Gilbert: When I think about where we are today, compared to where we were when we started this journey of virtualizing our network, we've come a long way. We've come a long way from virtually drawing up an architecture to time to market. We now have ECOMP and Domain 2.0 in production and we've achieved that milestone of time to market.

What we started doing in 2016 was really taking it from time to market to time to scale. That's sort of Phase II and that's where we've been focusing, taking it to that next level. That is going to take us to the next one or two years as we scale Domain 2.0 to 50% plus virtualization.

What we're now starting to look at is the third phase, which is time to globalization. Starting with time to market, then time to scale, then we're going to time to globalization. We're now providing ECOMP as the industry standard architecture to have other service providers, other service integrators, adopt worldwide. So that's where we are in the journey. (See Bell Canada Joins AT&T's ECOMP Roster.)

In the next three to five years, assuming we have this software-defined network, what are the value-added services that we could do? They will range from security all the way to cloud-based services, Internet of Things and drones that you couldn't have done before. What is that connected world going to look like? None of that is feasible or possible without this transformation of the network.

TT: In the short term, what is AT&T doing to meet its target of 75% virtualization of its network by 2020?

MG: We met our goal of 30% virtualization in 2016. We have a more aggressive goal for 2017, which is that time to scale that I was talking about. Once you get to 50% you're starting to see huge scale. That's sort of where we are ramping up to, but there are other elements to get us to the 75%. Some of them are internal. We talk about technology, people and processes and we'll continue to be in a rapid transformation in terms of the technology, trying technology and taking risks.

We're also looking at people in terms of skill sets and processes. We're looking at dismantling old processes and working towards newer processes with things like DevOps and continuous integration/continuous development, and open source. That's an important internal transformation to get us to the 75%.

We also talk about standardization with our vendor community. As we are virtualizing our network functions we want to make them look like Lego blocks that can plug and play into any service provider platform, as opposed to having the proprietary, legacy type of capabilities that we have today.

We need to get to that standardization where virtual network functions (VNFs) may be built differently but they look like Lego. They have simple APIs and they can be plugged and played into an AT&T network or any of the providers' networks. That's required to get us to get us to time to scale and time to globalization.

TT: Last year AT&T put ECOMP into open source with the Linux Foundation, can you talk about why AT&T made this move, and what it entails to create Open ECOMP? (See AT&T's Rice: ECOMP Reaches Critical Mass.)

MG: We've put a lot of things into open source. If you look back to AT&T Labs, AT&T Bell Labs, we put things like Unix and C++ into open source, but this (ECOMP) is a different beast. This is a significant, large project with a half million lines of code. It touches many businesses in AT&T, from consumer and mobility to large and small businesses. There's a lot of lessons learned that we've had in the past that we're applying here, but there are a lot of new things to learn as well.

We didn't develop ECOMP to put it into open source. We started it because our traffic was exploding to a point where many years from now we would not able to support the traffic evolution and exponential growth with the legacy technologies that we had. That's how it started. We didn't think "We have to have this in open source."

To get it into open source, there are the obvious tactical things like making sure the software goes through scanning, making sure it goes through single build, automated tests. So as complex as it may be, as many organizations that have been involved in this, which includes vendors, internals and organics, it had to come together as one platform.

At the end of the day, it's one platform. That's a major undertaking to take all of these different technologies, all of the different software, into a single build that passes security, that passes licensing, etcetera.

We wanted to make sure you could use this without having to pay significant licensing rights. It's open source. That's a major transformation to us. We've never done this at that caliber and complexity before. I call that tactical. We had to get to that point.

The second part, which is really difficult and challenging for any company, is the cultural transformation. A cultural transformation is not five people getting together and putting an open source out, or ten people getting together and extending something into open source.

This is the foundation of our network. It's the most complex project we've ever built. Hundreds and hundreds of people have been involved in that, internally and many, many more externally. So it's a cultural transformation to have the right skill sets. People that graduate from school don't know this stuff and we have to train our own people.

You also have to figure out that once you put it out how do you contribute to it constantly, and how do you take contributions from open source that are going to be sitting at the heart of our network? Different people will be writing software that's going to the heart of our network. So there's a major cultural transformation that we're going through as we prepare for Open ECOMP.

TT: What is the most important aspect of being a leader?

MG: There are a lot of books on this that people tell you to read. I went to business school and read a lot of books on this, but on the job what I've certainly learned is that there are probably three elements that are key to being a leader.

The first one is building, nurturing and hiring the best talent. You read it in books, you always hire and have the right people on the bus, but what's different here is it's not just building, nurturing and hiring the best people, it's also ensuring those best people work together like a team, like a perfect orchestra. That's really important and that's No. 1.

The second one is, just like with any job, is really expanding your technical knowledge while also making sure you have the business savvy view. You really need to combine both the technical side and business side together to be able to understand what is valuable to the industry and what is valuable to the company. You have to drive that down into what are the technological breakthroughs that are required? What are the technological extensions that are required, etc. Never get married to a technology or a tool because things change over time. That's the second one. The balance and the technical breadth that you need to have.

The third one is selling a vision and really having the team collaborate on the "what" and the "how" towards that vision. I've found those three to be very critical to the function that I hold today.

— Mike Robuck, Editor, Telco Transformation