Nan Chen has served as president of the Metro Ethernet Forum (MEF), a service provider consortium devoted to defining interoperability standards for Carrier Ethernet, since its founding in 2001. At the group's recent GEN14 conference near Washington, D.C., we sat down with Chen to discuss the Third Network, the MEF's ambitious vision for a flexible private network with Internet-like ubiquity and enterprise-grade service assurance.

What is the Third Network?

Chen: So, today you have two major data networks. One is the Internet, which has tremendous agility and ubiquity in terms of coverage. But it's lacking performance guarantees as well as having security issues.

Then you have another network, which is the Carrier Ethernet 2.0 network. Carriers built that for enterprise customers and internal use. That network is incredibly secure because it is privately built, and also it is performance guaranteed. One of the drawbacks, though, is it doesn't have agility like the Internet does.

I guess a question we were asking last year was, "Could there be a new type of network service would combine the best of the both worlds?" In other words, Internet-like agility and ubiquity and Carrier Ethernet 2.0-like service assurance, including security as well.

The answer we believe is yes, because the fact that in certain situations when you're not necessarily within the corporate office you may potentially need guaranteed service rather than internet access. Today there is really no other choice. You go to the Internet, IP VPN, or whatever's going back to your home office. We want to provide a choice to allow those willing to pay to get better service and build a private network for them. That's basically what the concept of Third Network is.

Available only for enterprises?

Chen: It's primarily built for end users, but the end users are employees of somebody. Potentially they are traveling on business, they are employees of somebody, and they're doing employee work. They should have the same SLAs as in the office.

If they're actually consumers at home, they may need guaranteed service for other reasons, such as high-quality video with family members, or kids and parents and so on. There is a possibility people could need a higher quality service.

How do SDN and NFV play a role in the vision for the Third Network?

Chen: SDN, on the highest level, is a separation between the control plane and the data plane, also separating the software from the hardware. NFV, on the other hand, is really virtualized network functions, originally provided on specific hardware.

Take an Ethernet circuit between Washington, D.C. and San Francisco. That service needs to be provisioned. With SDN and NFV, although they are going to be faster in terms of the initial configurations, they don't provide the capabilities to provision that circuit or automate provisioning of that circuit. That's where life-cycle service orchestration (LSO) comes in, which the MEF is defining in terms of capabilities and APIs.

LSO, SDN, and NFV—the three of them basically form three pillars for future networking, including the Third Network.

As it stands now, how does the vision for the Third Network intersect with other standards initiatives, like OpenFlow for controllers?

Chen: From an LSO perspective, we don't really care about OpenFlow, because that's actually [at a lower layer]. So we communicate with SDN controllers using OpenFlow, or some other protocols, to configure those devices.

There's been a proliferation of open source initiatives and standards-setting bodies. Do you feel that there's too many competing standards bodies, and how do you win out against all of these other groups that are trying to do similar things?

Chen: First of all, no one is really doing that LSO definition. We're the only standards organization doing that.

Secondly, on OpenDaylight, or OPNFV, or On.Lab — most of them are actually concerned about the infrastructure. We're really at a higher layer, in the service orchestration layer, which is an interface with the SDN controllers.

The carriers have a reputation for moving at a snail's pace when it comes to overhauling things in their network. Is this all a pipe dream? Is this something that could actually happen?

Chen: Yeah, it actually could. The reason being, if you look at the enablement of the Third Network — the reason is LSO. People realize if they want to migrate from existing infrastructure to the new infrastructure, they need to have the service orchestration abstraction layer. Which is what the MEF is defining, to be able to shield the application away from everything. That's one.

Number two is that transition, that LSO layer, would allow them to be able to gradually, at their pace, migrate to an SDN/NFV-based infrastructure without doing a forklift upgrade, so to speak. LSO really going to help them do that.

Number three, the Third Network initiative eventually will enable them to have a new revenue stream, which they are dying to have. With access fees of, say, $20, they could potentially aim that $20 to all the Internet access, whatever it is. Then on top of that they all collect usage fees. It's like the old telephone days — the model could be the same.

For many of your members, the carriers, for them to speak to each other anywhere else but within the context of the MEF might be collusion, or violating anti-trust laws. Is it cynical to think about the MEF as the legal venue for the industry to collude?

Chen: No, it's not so much collusion, but really promoting collaboration. Our bylaws really prevent us from talking about any kind of pricing discussion. It really is about a collaboration on a technical level — how to actually be interoperable to deliver that service end-to-end. Because there are almost no large enterprises that can be served by a single service provider anymore, so you need to have that collaboration between the carriers.

In today's telephone calls, you pick up the phone, you call somebody. That goes through so many carriers, and so automatically you don't even know about it. Seconds later, that call goes through and you talk and use that circuit. When you're done, you hang up. The circuit gets terminated, and the bill get settled between everybody.

That's where data service needs to go. The Internet can do it, connecting to multiple carriers, but the problem with the Internet is you can't really guarantee the service.

What we're really trying to do to basically build quality service like the telephony network has done from a control, management, employing perspective, but a high quality service for data.

What are the biggest challenges in building that?

Chen: First, we need to speed up the standard development. It can't take two years to develop, because typically that's how long it takes, 18 to 24 months. We wanted to be creative to make sure we got the standards developed faster.

Secondly is the adoption. It always takes a while to adopt, but I think we've done it before in terms of Ethernet. I remember at the beginning of the life of the MEF, Ethernet was taking a while to adopt, because they said, "You'll cannibalize my TDM service."

Then finally people realized you would rather eat your own young. You know? Instead of having someone else eat it.

When we get back together for next year's conference, where do you predict development of the Third Network will be?

We certainly would have the LSO initial specs. Probably the initial code base generated and initial implementations. We may start talking about certification of LSO and then potentially a Third Network service. And also we see collaboration between our organization with the ONF or ETSI or OpenDaylight and OPNFV — really helping to make what we call the MEF Unite program, which is a way to get everybody together to deliver the Third Network.