Carrier Ethernet equipment vendors generally fall into two camps these days. Most carriers would prefer to camp somewhere else

“ The Way Carrier Ethernet equipment is sold today, we tend to see vendors in one of two camps: the Sonet-replacement, claw-your-way-up-from-the-bottom-of-the-network guys or the enterprise-services, claw-your-way-down-from-the-top guys. But operators I've talked to aren't particularly interested in either one of those two guys. They're looking for a third guy — a guy who says Ethernet is to metro infrastructure what IP is to the core. A guy who will show me I can create services and do traffic engineering in a flexible way on top of Ethernet. I don't care if it uses [provider backbone transport] or [provider backbone bridges]; I just care that I have those sets of properties. Because there are situations in my network where I believe deploying IP is not my best choice.

“You could argue that the logical star topology of metro networks tends to favor simple Level 2 solutions such as Ethernet. Almost all the traffic on metro networks is focused on getting out of the metro. In core networks, you have a relatively promiscuous distribution of traffic among the exit points; there's no real predictable path. So connectionless IP protocols are great for core missions. But in the metro, if all my traffic is going to an off-ramp, what do I need routing for?

“In the U.K., the higher density of demand makes it possible to extend metro architectures between metros. BT became a focus for IP/MPLS versus Ethernet PBT. It had a choice between deploying Level 2 or 3 across a wider area. Most operators don't. They'll do Ethernet or IP, and all they're trying to figure out is how to optimize whichever choice they make.

“There are other things going on at the demand level that are beginning to make metro infrastructure look a little more like the core, and vice versa. So over time it's very likely there will be increased collision between the PBT folks and the MPLS folks and between the Ethernet guys and the router guys. (I've never had clients get mad at me for saying things like, ‘Your company's going to die,’ but I've had people get mad at me for saying, ‘You should consider PBT.’ That's really offensive.)

“Meanwhile, operators are saying, ‘I have traffic growth. I need a metro architecture that, first and foremost, supports my current incremental service missions, the ones my old architecture can't support. I'd then like it to gracefully absorb traffic from other missions as the equipment associated with those other missions ages out of rational sustaining. If I have users taking TDM voice channels on Sonet pipes, for example, why would I want to force them to convert to some kind of TDM over Ethernet?’

“When someone comes in and says, ‘I have the perfect answer to your resilient packet ring problem,’ they say, ‘I don't have one.’ If someone says, ‘I have the perfect answer to your carrier Ethernet access problem,’ they say, ‘I don't have that either.’ They have a metro bandwidth problem. The irony is carrier Ethernet and PBT solve that problem rather well. But there are still issues with carrier Ethernet, and those issues are predominantly linked to the fact that the carrier Ethernet process has been created by these two disparate camps. Because they haven't come together and created this third guy who's got the infrastructure mission, the voids associated with Ethernet in an infrastructure mission have never really been uncovered.”

TOM NOLLE is president of CIMI Corp.