In the early days of SD-WAN, end users needed to deploy boxes on premises. That is changing, according to Jeff Lewis, Comcast Business's vice president of connectivity services. Companies still can locate boxes on premises, but they have a variety of off-site options.
In the second of a two-part conversation with Telco Transformation, Lewis said that the goal is integrated platforms that can more easily and elegantly provide VNFs to end users. (See Comcast's Lewis: No MPLS Customer Base an Advantage for SD-WAN Service.)
Telco Transformation: How can a network operator future-proof against the near-certainty that many SD-WAN vendors will go away or be acquired in the near future?
Jeff Lewis: The way we look at the vendor community today is relatively straightforward. When we entered into this conversation, we insisted that our solution needed to be a long-run platform that would evolve and is not a "bespoke" one-off approach. Our view of the world is that orchestration is paramount. SD-WAN is not the first or the last product that we're going to offer in this effort.
When we talk SD-WAN, maybe we're using the term incorrectly, according to the lexicon police. We look at SDN as an extensible, orchestrated platform where SD-WAN is the first hosted VNF. Our SD-WAN product offers a virtual router capability, which unto itself can be a VNF, and it comes with a stateful firewall, which also can be a VNF. They all come together in our SD-WAN offer. We're going to continue to offer incremental VNFs through that harmonized, orchestrated solution. It is hosted in our network. The controls and directors live in our world.
We and our partners have gotten the architecture right. That architecture is going to allow us to spin up more and more VNFs. I'm going to stay away from vendor names but that litany of [SD-WAN] vendors, every single one of them, at some juncture potentially could be a partner of our ecosystem where we will be able to interact with that product for our customers. And that could include, and I am by no means suggesting that we will do this, a separate SD-WAN partner.
What we see today and what is challenging about the market is that the first to market were standalone box vendors. The VeloClouds, the Viptelas, the CloudGenixes, all of those guys all were box-based. The exceptions were some of the smaller companies, such as Aryaka, which have a hosted approach. This is how it went: "Hey, Mr. XYZ Company, I'm really interested in SD-WAN." "Great. Let me sell you my box," which is analogous to “Let me sell you my IP-PBX.” They would put it on premise and then the vendor would say, "Oh, Mr. Company, you now need to run that box. You need to go get all the software upgrades yada, yada, yada."
You need to worry about getting your broadband links and all that stuff. You go manage the network, but it is a single piece. I call it bespoke. It goes in your data center and you're done.
Then if you want to do the next VNF, let's say universal threat management, some of these companies have found a way to integrate with a third-party VNF. Terrific. Now you have two, but there is nothing else beyond that. There is no integration, no harmonized digital experience. There is no integration into the network. There is no contemplation of orchestration for the next VNF. How am I going to do that? You put a lot of pressure on the end user where they're ultimately going back to the world where you're going to stack stuff up in their data centers as opposed to really virtualizing.
From the vendor community we only saw two vendors and one of them is the one we felt was the leader at the gate, Versa Networks , which could scale. We didn't think the other guys could scale in a carrier class, multi-tenant environment because we made the choice to host.
Track the heartbeat of the virtualization movement with Light Reading at the NFV & Carrier SDN event in Denver. There's still time to register for this exclusive opportunity to learn from and network with industry experts -- communications service providers get in free!
TT: How does Comcast Business think this will play out in the long run?
JL: We have a very long-term view of SDN overall. We have a view of full orchestration across the entirety of products and services where you look at a box, which could be an elegant, sharp device, but to us it felt like it was just a silo. It was not integrated into a carrier solution.
Ultimately, our customers are going to be faced with an opex and a capex question: "Do I want to buy this box and buy the licenses and buy, buy, buy and then get all the support, support, support?" That's one path that they're all familiar with. They could choose that and that's great.
They also can go the path of, "Let me go find a managed service provider who is going to aggregate all this stuff around me, but still basically just give me the effect of SD-WAN, that single service experience. Because the managed service provider is not going to orchestrate in any way, but they'll help and hopefully give a good service experience for their customer. Or you come down the path that we're going down which is, "I'm going to fully host this bad boy. I'm going to give you a world class digital experience, and I'm going to only put an appropriately sized universal CPE [on premises].”
TT: Will everything end up being in the cloud?
JL: Some VNFs can be hosted in the cloud. Some will work well in the cloud, but some really only make sense to live locally on the customer's premise. You do need some point of presence from an SDN control perspective and that's the only thing the customer is going to see on premise. So the customer is going to be offered this option of either I can own it and control it because that's what I've been doing for 15 years. I've got to do it. Or they can turn to a carrier like us and they can get that hosted experience, but be given effectively complete control of the entire service experience.
— Carl Weinschenk, Contributing Writer, Telco Transformation