METRO NETWORK REROUTE

As longtime telephony carriers evolve into triple-play providers, much attention has been paid to the architectures and technologies in the access network — and at the home — that will make this transformation possible. But scaling those networks has a massive impact much further upstream in the metro network. And so metro networks are being redesigned and redeployed to accommodate video traffic.

Obviously, the addition of video content to broadband service drives the need for much more capacity in metro networks. Suddenly, carriers have to deploy numerous 10 Gb/s links to move video from headend hubs out to neighborhoods. In recent months, equipment vendors have cheerfully reported broadband already stretching the limits of metro capacities. And with high-definition television, video-on-demand and gigabit-speed fiber to the home all advancing fast, those vendors ain't seen nothing yet. But a move to bigger pipes is just one of the changes necessary to transform today's metro voice and data networks into video distribution machines.

Verizon Communications perhaps best articulated how ill-suited today's former Bell company networks are for video by using its existing network to deliver its first video services in Texas. The setup there is essentially a wavelength division multiplexing (WDM) ring with lots of stacked, subtending OC-192s. Broadcast and on-demand video run along these rings and drop off at what Verizon calls “video service offices” (video nodes that serve several customers each), which contain stacks and stacks of add/drop multiplexers (ADMs). This architecture does the job; it serves live customers today. But to Verizon, it's a living example of how not to build a metro triple-play network.

“It's an ugly network,” said Stuart Elby, Verizon's vice president of network architecture. “You have this monstrous mechanism of rings. It doesn't work. It clearly is not going to scale to 35 million users.”

To build the next-generation triple-play metro network requires the next generation of equipment, which carriers will deploy in bulk this year. One of the pillars of the new metro architecture is the reconfigurable optical ADM (ROADM), which allows carriers to switch and provision wavelengths — a task that has historically required an expensive truck roll and a manual adjustment — simply by clicking a mouse. AT&T will deploy 400 ROADMs this year to enable IPTV delivery, for example. With ROADMs, the carrier will be able to transport wavelengths rather than Sonet STS-1 circuits, carrying more bandwidth and cutting down on expensive optical-electronic conversions in distribution rings.

“You hear a lot about servers and set-tops, but ROADMs are the enabling technology that really makes IPTV possible,” said G. Keith Cambron, senior vice president of AT&T Labs.

And because metro ROADMs dissolve the need for manual provisioning, they make manageable what would otherwise be exorbitant costs for operating a video-bearing metro network. To deliver IPTV without ROADMs in the city of Los Angeles alone, AT&T estimated it would need to hire 30,000 more workers.

Because of these benefits, metro ROADMs have long been in demand for interoffice networks. But the addition of video requires metro ROADMs to be even stronger. A typical interoffice ROADM ring today includes about four or five nodes, but a network distributing bit-rich video to the masses might require dozens of nodes. Therefore, the optical gear comprising those rings needs to have greater reach because although it may not need to send signals over a greater physical distance, it will need to send them through far more nodes.

“That means your optical reach — or budget, if you will — is much different and starts looking more like regional or long-haul transport rather than simply metro transport,” Elby said.

Last year, Verizon issued a request for proposals for next-generation ROADMs to meet rising bandwidth needs and simplify metro networks at the same time. The gear they wanted would need to be protean and versatile, acting as a ROADM where needed or a regular Sonet ADM as the case may be. Verizon also wanted pluggable optics and something the company calls “ADM on a wavelength” — 10 Gb/s ADM functionality on a blade that allows traffic to be dropped as a single wavelength without the need for external jumpers. One of the aims, as Elby put it, was, “to take advantage of optics where optics is good — broadcast [video] and pass-through — and use electronics to do what electronics does well.”

“This product is designed to evolve from something simpler to something more complex,” said Dana Cooperson, director of optical networks research for Ovum-RHK. “Call it a ‘supercharged ROADM.’ In theory, this does it all.”

Fujitsu Network Communications was favored to get picked for this task, having dominated the metro ROADM market the year before (hogging 80% of the domestic revenue). But Tellabs is reported to have snagged the deal in a late upset that gave the vendor a jolt of vaunted credibility in the nascent next-gen transport space.

Verizon conducted a study to determine what a metro network would be like if it were built with such next-generation ROADMs rather than the stacked Sonet rings and ADMs of the carrier's existing repertoire. The result was a substantial improvement on several levels. The new architecture would reduce the total number of network elements in a chosen area from 160 to 35. Whereas the old network's equipment occupied more than 100 bays, the new one would take just 32. The number of fiber pairs would go from 20 to six. The new network would consume more than 60% less power and save Verizon 50% to 60% in capital. And whereas the old network was near its bandwidth capacity ceiling, the new one would have room for growth.

Verizon is set to install these next-gen ROADMs sometime this year. It will use ROADMs to connect video hub offices to large distribution rings. Video (both broadcast and on-demand) will swing around a common ring with Internet data until they alight on a drop (optical or not), where they will head out to a video service office and toward the customer.

“It's still a complex network, but it's much simpler and much less costly than the existing technology in the network,” Elby said.

The next step, he said, would be ADMs and switches that route and groom traffic in its native packet form. “Vendors aren't there yet, but that's where we need to go.”

Networks that carry video are also inherently different from phone and broadband networks in another important respect: Whereas voice networks were built on the assumption of bi-directional traffic symmetry, video is pretty much a one-way street. (Residential broadband has been characteristically asymmetrical, but the ratios of incoming to outgoing bits is nowhere near that for video.) This is highly frustrating for carriers because the same video traffic that forces them to put really big, expensive pipes in the metro also dictates that much of those big, expensive pipes goes to waste.

“I need [to deploy] 10 Gb/s in both directions, but I'm only using the interface in one direction,” Elby said. “That's very costly to me. I'm looking for unidirectional optics, so I can get 10 Gb/s going down and don't need to pay for the laser coming back up.”

Eventually, carrier networks will become less symmetrical as a cost-saving measure. But carriers (and therefore vendors) aren't in a big rush to make that happen, as they've already paid for these symmetrical networks, and it's always cheaper to use what you already have than it is to buy something new.

“Even though you're not necessarily using that channel in the other direction, there isn't anything else you could really do with it,” said Glenn Wellbrock, Verizon's director of advanced network technology. “It's there — you bought it. You've got amplifiers and ROADMs in place. If you were to build a one-way system, how much money would you really save?”

Over time, these networks will also evolve more toward mesh-based topologies and away from their current ring-based structures. And the old distinctions between metro and regional networks — which is already blurring — will fade. Eventually, the networks that are already beginning to carry video traffic today might actually appear to have been designed for that purpose. And the ugly network will be a relic of the past.

VERIZON'S NEXT-GEN METRO STUDY

Impact of architecture on a chosen metro network

	OLD NETWORK	PROPOSED NETWORK
Total network elements	160	35
Equipment space (in bays)	108	32
Fiber pairs	20	6
Power consumption (kilowatts)	80	30
Capital savings	—	50% to 60%
Bandwidth capacity	Near exhaustion	Room for growth
Source: Verizon