EFFICIENCY MEANS STAYING ON-NET

Steady growth in local data traffic termination has taxed metropolitan networks that weren’t originally designed to handle the accumulated transmission needs. The bottlenecks create a quandary for those that move metro traffic: Improving efficiency can lower network and personnel expenses, but the cost of that efficiency is often prohibitive.

It's no secret, given how the public network was built, that the way metro traffic flows is often less than efficient. For example, an e-mail that originates and terminates in Minneapolis may first travel to a central office in St. Paul, Minn.

“It costs a fortune to pay for long-haul capacity to get up and back,” said Hugh Martin, CEO of ONI Systems, a metro optical equipment vendor. “It's much more efficient to keep local traffic local.”

That's because keeping traffic local, or “on-net,” is strongly tied to lowering costs by improving a provider's network efficiency and its ability to strengthen service level agreements (SLAs).

The problem for many providers is that while it sounds like a good idea to keep traffic on-net because of the potential savings in cost and network capacity, that might only be achievable by spending exorbitantly.

But providers are at least beginning to chisel away at the efficiency issue. How they free up congestion and how much they spend to do so is as varied as service provider networks themselves.

In many cases, the providers in most need of improving network efficiency are those with the capital to do it: the incumbents. Yet many incumbents are feeling less pressure from their increasingly weakening competitors. And if service quality doesn't vary much from one service provider to another, pressure from customers is likely to slump as well. That, in turn, is slowing the network upgrade process.

“The ones with the most money to spend are the ILECs,” Martin said. But most ILECs are around six months away from signing deals to upgrade their networks.

About 50% to 55% of traffic on metro networks stays metro, according to Martin. That number is likely to rise as traffic increases, which will cause service providers to sniff out the ways to boost efficiency with existing assets. “It's a market that hasn't been tapped yet. Local traffic hasn't taken off because we don't have enough connectivity,” he said.

To lessen the need for customer traffic to jump onto competitors' networks, providers such as WorldCom and Level 3 Communications have implemented programs to attempt to keep more traffic on-net.

“We offer different pricing depending on where [traffic] goes or comes from,” said Jack Waters, chief technology officer for Level 3.

The provider offers a few different pricing levels with its (3) CrossRoads offering, where the cheapest costs are achieved by staying on Level 3's network from origination to termination. The idea behind the offering is to increase network efficiency, lower the number of handoffs and fulfill SLAs. Concurrently, the provider saves money and bandwidth.

Service providers also have to keep their networks as clean as possible, according to Jay Adelson, chief technology officer for Equinix, an interconnection service provider. The only way to accomplish that is to stay on-net, he said.

“Peering is a big issue. If the link between Level 3 and WorldCom [for example] is congested, how can they make guarantees? They can't,” Adelson said. Therefore, carriers should try to find ways to circumvent or avoid traffic handoffs, he said.

Although service provider networks vary greatly, they all have similar choices when it comes to the network configurations that will help them meet demand and improve efficiency.

There are basically three different options, according to Martin. One, providers can use some version of dark fiber to run links to key customers. Two, they can continue to scale and build by replacing Sonet rings and adding rings on top of rings. Three, providers can use an optical add/drop multiplexer architecture (see figures).

And while many—even staunch Sonet supporters—would agree that the latter approach is the most efficient, it is also the most expensive.

With times being tight, providers are increasingly choosing to run dark fiber or add Sonet rings, according to Martin. And while that is painful in the short term for metro DWDM players such as ONI because of lost sales, providers do see the value in opting for metro DWDM for added efficiency.

But providers need to be able to do more than just set up a 2.5 Gb/s line and then tear it down, according to Rob Lane, vice president of business development for Marconi. They need to have the flexibility to provide bandwidth on demand to further improve network efficiency.

Of course, carriers aren't necessarily betting on the importance of flexibility — they just want to make money. But if more flexibility is a catalyst to revenue, so be it. When they run dark fiber or build Sonet rings, “they know it stinks to have no SLAs, but it may be the only thing they can afford,” Martin said. “It really just depends on the cash a company has. Carriers are identified by providing end-to-end solutions.”

That has caused providers to opt for things such as wavelength leasing, which essentially provides them access to metro capacity that they can treat as their own. Instead of blanketing its facilities with DWDM equipment, Level 3 is relying on its own fiber and partnerships.

“We are purchasing local dark fiber where the economics afford us the opportunity to interconnect our offices and enhance our ability to rapidly provision onto our backbone,” said Jack Norris, head of customer service and network for Equant. “We use Sonet/SDH as required but are looking on an IXC basis to connect equipment directly to wavelengths.”

So far, though, the provider has only bee able to make the direct connections on routers and not on ATM equipment. As a result, the provider needs Sonet/SDH to derive OC-48c speeds to interface with its ATM platform, according to Norris.

Other carriers such as Broadwing and Qwest Communications are doing the same.

“As we move forward, services require changes, and bandwidth requires change, too,” said Derek Hodovance, vice president of IP architecture for Qwest, adding that the carrier is watching the Ethernet LEC carefully. “We definitely want to eliminate [or lessen] the number of devices.”

By doing so, the provider would significantly lower its equipment and operational costs and the need for so many network elements to tie together.

Qwest is now concentrating on further developing its metro areas.

As for Level 3, “we are focusing on pulling a large cross-section of fiber and conduit in metro areas, which makes us less fiber-constrained,” said Waters, noting that the provider will use metro DWDM in areas where fiber is maxed out. “We wouldn't want to use it in all cases, though.”

While keeping traffic localized and on-net are good efficiency strategies, some suggest preventing traffic that could be kept at the metro edge from creeping into the metro core also is critical. Essentially, by consolidating traffic at the metro edge rather than the core, carriers can gain more efficiency in their networks.

“Only traffic that needs to be backhauled [goes to the core],” Lane said. “And reducing the traffic in the backbone allows scale.”

While there is some validity to the argument to keep traffic at the edge, “reducing the amount of traffic going into the core isn't really needed,” said Sam Greenholtz, senior analyst for Communications Industry Researchers. “Providers aren't going to spend additional money to do it.”

While some may not spend money on that, it behooves them to analyze and streamline how traffic traverses their networks. When traffic arrives in a CO, for example, it hops on the first circuit available, regardless of where it's headed.

But when large, sprawling networks are concerned, sometimes simply identifying the source of inefficiency problems is as much of a headache as fixing them. “It's difficult to determine where each [piece of traffic] is going,” Greenholtz said. “It's difficult to do, especially with cutbacks in personnel. They don't have the time to do that, so it just goes on the first available line.”

That means the networks are running inefficiently, said Mark Lutkowitz, vice president for CIR. The reality is, they automatically end up with inefficient transmissions, he said.

And although most carriers embrace the idea of improving network efficiency in theory, keeping traffic on-net doesn't always work in practice. That's because for customers, it means putting all their faith in one provider.

The more sophisticated customers become, the less likely they are to go to a single source for their services, he said.

“Carriers have come to play on a level playing field, and they can't hold customers hostage anymore,” Adelson said.

Others seem to agree that despite providers' best efforts to keep more traffic from hopping off their networks, those efforts may be futile.

“[Providers] may say they get better quality of service, but it's really not quality but higher margins they are looking for,” said David Gross, senior analyst for CIR.

“If you follow the dollars, in the end, providers like Level 3 and AT&T make money by packing the glass,” Adelson said. “They are very motivated by getting people to use the fiber.”

So does that mean providers will admit there is room for change or that they need it? Not necessarily.

“We continue to make streamlining decisions,” Level 3's Waters said. “Equipment does come at a cost.”

Service providers aren't rushing out and building ideal networks, but they do have opportunities to build on what's there. “They are going to look for the cheapest way possible,” Greenholtz said.

But even though there may be room for improvement, the perfectly efficient network may be a reality relegated to the labs.