To hear some people talk, you'd think copper cabling at the WAN level was passé, being replaced by fibre-optics. To them, fibre is where it's at - even outside the office, where they're boosting FTTH, or fibre-to-the-home.

Yet where are the teams of workers digging up the streets to lay all that fibre? They're not there, and neither is the fibre. The truth is that most commercial and residential sites are still connected by copper - or if you're the unlucky inheritor of certain past cost-cutting measures, aluminium.

That's why at the recent NetEvents forum, quite a few of the high-level discussions involved new ways of using that copper plant. In particular, that's ways to run long-range Ethernet connections over copper - such as a project called Carrier Ethernet, which has been co-ordinated by the Metro Ethernet Forum (MEF).

"Ethernet services are becoming mainstream," says Chip Redden, VP of marketing and product management at Overture Networks. "Everyone's focused on fibre and talks about copper being legacy, but there's a heck of a lot of copper out there and it ain't going away any time soon, so it's got to be used.

"For example, only eight percent of European office buildings with more than eight people have fibre, and in the US it's only 14 percent. All those remote offices too - most will be on copper. So you have to have a solution for copper - plus, a lot of service technicians know how to provision and test a T1."

Emulating TDM

So Overture has developed technology to bond several E1/T1 links together, run Ethernet over them. Like several other developers, it then uses circuit emulation to create a legacy TDM link for voice telephony over the packet network.

This is another area where the MEF has been active, says Olle Gustafsson. He is standardisation director at Ericsson Broadband Systems, but he also works with the MEF, and says it has developed "a test suite to prove that the IP circuit and equipment can do [TDM] circuit emulation."

He adds that there is still a clear need for TDM services, but as a network technology it's going to lose out to Ethernet and IP.

"This issue is E1/T1 scalability - Ethernet is more scalable. But the operators wanted to be sure Ethernet was as reliable as E1/T1," he says. "It's the same decision that enterprises took in deciding that Ethernet was ready and mature enough for mission-critical applications."

It would probably have been possible to update TDM beyond the limits of E1 and T1 - that's 2Mbit/s and 1.5Mbit/s respectively. Gustafsson argues though that it is easier to simply move over to Ethernet - not least because Ethernet is a much bigger market, so it offers economies of scale that an uprated TDM would not have.

No more subnets

With most of these long-distance Ethernet schemes there is no longer any need to have different subnets for each office, routing between them at Layer 3. Instead, they all become part of the same Layer 2 network.

That's part of what the MEF was trying to do in standardising Carrier Ethernet, says Kevin Vacheron, the forum's chief operating officer.

"It is about adding attributes to Ethernet to make it carrier-class," he explains. "What matters is that it performs like Ethernet end-to-end, not what's underneath. Then there's no need to buy conversion equipment, because you're taking layers of conversion and complexity out."

Chip Redden agrees: "Carrier Ethernet means you finally have a label for what the carriers are trying to do - it's a common language for carriers and vendors," he says.

Back to the copper though, and if what matters is that you see Ethernet at each end of a line, is there any need to keep E1/T1 technology in there at all? Not really, argues Kevin Hatteras, whose eponymous company has developed a different - and simpler, he says - route to the same destination.

"Rather than run Ethernet over an E1/T1, we take the raw copper wire and run Ethernet on it - you can lease a dry copper pair [in the US] for $12 a month, then run Ethernet on it," he explains.

"The sweet-spot is 10Mbit/s to 20Mbit/s symmetrical - it's not like ADSL, more like SDSL, but carrier-grade and it's Ethernet on the wire. We can go two miles at full speed, or three miles at reasonable speed. That's better range than ADSL for the speed, and it is getting faster and faster, with all the improvements in digital signal processing."

He says that Hatteras Networks' main target is mid-sized enterprises, who are willing to pay around $1500 a month for such a service, which he calls "Mid-band Ethernet", and which follows the IEEE 802.3ah specification for Ethernet in the First Mile, or EFM.

"European pricing is very different by country," he adds. "Typically they peg it off the E1 price, so a 10Mbit/s circuit costs more than one E1 but less than five E1s." The equipment costs vary similarly, he says - more than the equipment for one E1 and less than the equipment for five.

The mobile data 'bump'

It's not just offices that are driving the need for metro or first-mile Ethernet though - it is mobile networks, and especially the need for more and more backhaul capacity to cellular base stations.

Even more specifically, says Olle Gustafsson, that mobile backhaul traffic is shifting from being TDM-centric to being IP-centric, as the network operators finally figure out how to sell mobile data services.

The problem is that in many cases they seem to have done it without adequate planning. It is not clear whether that was down to a lack of foresight, a shortage of money after investing in 3G, or a combination of the two.

But data volumes are rocketing at last, and many cell sites don't have the backhaul capacity to cope with the rate of growth - and that's going to provide a healthy boost for mid-band Ethernet, says Kevin Hatteras.

"We're now getting pulled into mobile backhaul - the operators are needing to upgrade fast to serve data," he explains, adding that "it's typically two to three T1s per base station, serving three or four carriers," which means about 1Mbit/s per carrier.

And as more and more of us move over to mobile broadband, that's going to make the idea of running 10 or 20Mbit/s over the same copper start to look awfully attractive.

Find your next job with techworld jobs