CenturyLink is in an NFV state of mind, and carriers that seek to dive into those same waters need to make sure their employees are all on the same page as well, according to the telco's Bill Walker.
In the final installment of this Q&A with Walker, director of network and cloud architecture for CenturyLink Inc. (NYSE: CTL), he spoke about the impact of open source communities on NFV and the need for carrier employees to have the same mindset across the board. In the previous Q&As, Walker discussed automation and the history of CenturyLink's NFV journey as well as offered his advice for service providers that were just starting down the NFV pathway. (See CenturyLink's Walker Talks Automation, Orchestration and CenturyLink's Walker on NFV: Just Do IT.)
Telco Transformation: NFV seemed to hit a bit of a lull last year. Do you agree and was that true for CenturyLink?
Bill Walker: We hit a lull too, but our lull was more of a development cycle. I think the standards in open sources are starting to coalesce a little better. They are starting to congeal a little better. Obviously we're not a big OPNFV shop. We're looking at ONAP and some of the pieces that came out of ECOMP.
TT: Since you mentioned ONAP, what are your thoughts on open source?
BW: We're going to [install] an OpenStack shop. We've been an OpenDaylight and ONOS shop all along. The open source stuff is really pushing forward and is fairly strong. What we haven't seen is the management standards take hold. ECOMP, or ONAP, is still very new. We're very much in support of having more standard management frameworks around it rather than just the bits and tools underneath.
OpenStack has been solid for a really long time but now how does OpenStack actually interface up to talk inventory? ONAP and ECOMP and some of the OPNFV work have really focused around those management interfaces.
Obviously in 2013 those didn't exist when we started with NFV, so it was very difficult to write around. We're more than happy to share with the industry as we go forward just so we can all get there faster and not all solve the same problem.
TT: Do you feel as though the traditional hardware vendors are doing enough for NFV, or is it more about startups now?
BW: We actually work with a lot of hardware vendors who are asking for our input on how to better serve NFV and not just about servers and storage. When I say, "I want to go all SSD in a server," and they ask me why, I tell them I don't want a little electric motor and some bearings and rotating parts in an environment that has heat and humidity issues or a different operating environment than they're used to. We are also seeing a lot of network equipment providers -- traditional NEPs -- who are now engineering servers as well. If you go back to the open-source side, a lot of the stuff coming out of OCP has been having a big impact in the hardware market.
Sliding back even into the ODMs, they're really contributing in environmental space airflow, and even the power to do more friendly central office and metro office deployments.
TT: One of the sessions in the upcoming Light Reading NFV and Carrier SDN event is about carrier grade NFV. Do you think we have carrier grade NFV, or is there more that needs to be done?
BW: We do, but it's not easy to do. There's a bunch of people with IT skills out there in the market and then there's a bunch of people with 30 years of telco experience. Mixing the two and finding the common language that they can design and architect together is difficult. And forklifting software off of a black box and making it run in a VM doesn't help us. A lot of the VNFs that we're seeing now – which have started to evolve over the past year, year and a half -- are actually designed with what we call cloud native. It's not software that runs in a hardened box. It's software that's designed to fail over, and maybe load balance, and spread some of the risk around. It is possible to create carrier grade services, but not if you think an x86 server running at an unfriendly central office in a warmer than usual environment is going to survive for five nines (99.999% reliability).
TT: Any final thoughts related to NFV?
BW: My biggest thing is there's a way (to do NFV) -- open the minds and look at the skills and not lock yourself into a traditional network mindset. NFV is a big trigger for us to do that, but you still have to follow through.
TT: But part of that is getting the right employees. You mentioned there's these IT employees out there that maybe service providers need more of, or is it a matter of reskilling current employees that have done telco forever into some of these new areas?
BW: It's not just training. The biggest problem is communication. I can send you off to training and if you come back and you're just still thinking the old mindset, then it's broken. It's about communication and actually maintaining the change. It's about having the networks guys and the IT guys having steady continuous conversation to find that middle ground where they can understand each other and the terminology all fits.
You can go off and get some VMware OpenStack training for a week, and you're by no means an expert. But if you sit down and actually hold conversations with people and ask the questions, then you start to behave differently. You start to think differently. Having the 30 years of the pure network mentality in the room with the IT guys -- who just figure stuff breaks and we'll fix it tomorrow -- you come to an understanding of what the demands really are and how hard it is. But that's how you find the solutions that span both.
— Mike Robuck, Editor, Telco Transformation