METRO DWDM: IS IT THE RIGHT TOOL FOR THE JOB?

Feb 1st, 1999

METRO DWDM: IS IT THE RIGHT TOOL FOR THE JOB?

BY STEPHEN HARDY

Judging from the latest generation of equipment--and the marketing artillery that manufacturers have redeployed from the long-haul front--the next target for dense wavelength-division multiplexing (DWDM) is the metropolitan marketplace. Increasing bandwidth demands are as much a problem for incumbent and competitive local exchange carriers (ILECs and CLECs) as they are for long-haul carriers, the story goes. Fiber has now become a necessity for these carriers. But building additional capacity into densely populated metropolitan areas is expensive and time-consuming. DWDM holds the answer for fiber exhaust today and enhanced service provision tomorrow, the pitchmen say. It works in the long haul, and it will work for you--just sign here.

Local service providers can be forgiven for scratching their heads while pondering such blandishments. Metropolitan area networks have different design and economic underpinnings than long-haul networks. So how can local carriers decide if what is good for the goose is indeed good for the gander? One network design consultant who lists both local and long-haul carriers on his client roster has been telling his customers the time is right to at least consider DWDM for their metropolitan area networks. However, there are several factors to weigh before making a decision that, if handled improperly, could strand your expensive DWDM capital investment in a few short years.

Worth a look

The notion of DWDM in metropolitan area networks is more than a fiendish marketing ploy, according to network design consultant George Vinall, president of International Protocol (Annandale, VA). Vinall`s clients include CLECs like Next Link and WinStar as well as long-haul carriers like Qwest Communications International. The metropolitan DWDM application interests both carrier types. The reason behind the CLEC interest is obvious. And Qwest is keeping an eye on this issue because it needs to interconnect its DWDM-based long-haul network with metro networks in the cities it serves. Vinall has told clients of both stripes that several trends have aligned to make the technology attractive to local carriers.

"I think the biggest trend is that the cost of the equipment is going down to the point at which it becomes a much more attractive alternative for carriers than it was in the past," he says. Previously, DWDM equip ment had been scaled for long-haul applications--and long-haul wallets. Most of the major DWDM equipment vendors have announced smaller, less expensive complements to their long-haul product lines directed toward metropolitan applications. Many of these lines should be available this year.

These lower-cost products come not a moment too soon, as the fiber-exhaust problems that led to the acceptance of DWDM in the long haul have begun to crop up in ILEC and CLEC networks. "In general, local-exchange providers are running out of capacity and don`t want to lay new fiber, because it`s so expensive to lay new fiber," says Kathy Lynch, senior product manager for Qwest`s Carrier Group and one of Vinall`s co-strategists at the communications company. "So, as a result of that, it has become more economical to deploy dense wavelength-division multiplexing." Lynch says fiber exhaust is most acute within ILEC networks; thus, the regional Bell operating companies (RBOCs) will serve as the initial drivers of the metropolitan DWDM market.

The desire for increased control of the network also has made DWDM appealing, says Vinall. Here, the RBOCs drive the market as well--although not in a manner they`re likely to brag about. "People have suffered at the hands of the Bell system for so long in terms of loop deployment and the coordinated conversions that the metropolitan customers are very leery of having any other technicians deal with their equipment," he asserts. It`s not that the Bell companies are inherently incompetent--they`re just caught in the technological undertow. "The digital revolution hit them as being as big a surprise as everybody else, and they have very few technicians even today who`ve been trained on optical technology to the point where they can just slap a loop up and test it and get it working in no time. It`s a real problem for them," Vinall states, adding, "Of course, it`s a real problem for everyone else, too."

DWDM equipment--and optical networking in general--will give CLECs the ability to dole out bandwidth themselves to their customers with the granularity necessary to provide next-generation services, particularly as the promised ability to add and drop wavelengths to customer sites becomes a reality. Naturally, the more sites such equipment allows carriers to service simultaneously, the wider the cost of the equipment can be spread across the customer base. Simultaneously, the ability of DWDM equipment to accommodate a variety of transmission schemes will enable Internet service providers (ISPs) or other data-centric companies to complement their current service portfolios with the addition of voice or other offerings using Asynchronous Transfer Mode (ATM), frame relay, or Synchronous Optical Network (SONET).

Yes, butU

While DWDM promises several benefits to the metropolitan market, carriers should remember that the technology still has its shortcomings, say Vinall and Lynch. For example, while equipment prices are falling, DWDM technology remains expensive. Transponders can cost as much as $30,000 each, while amplifiers also can cost $30,000 if they can`t be acquired as part of a package deal. Don`t think you`ll need amplifiers because you`re not running a cross-country link? Vinall points out that metropolitan area networks usually involve significantly more add/drop points than long-haul links. Each of those points introduces a loss penalty to the network. While admitting that DWDM vendors would argue the point, Vinall believes that given enough add/drop locations, certain metro network segments will require some sort of amplification.

Regardless of whether amplifiers will play a significant role in metropolitan area networks, DWDM equipment in general must continue to drop in price before the average local carrier will feel comfortable with the economics, Vinall and Lynch predict. Vendors tell Vinall that they aim to make metropolitan area equipment such as transponders approximately one-third the price of their long-haul counterparts.

Another short-term problem with DWDM resides in the fact that the equipment to back up some of the technology`s promises, particularly in the area of wavelength-based services, is neither mature nor widely available. Optical add/drop multiplexers (ADMs) are still in their infancy, with 4-wavelength systems the current standard. "Really, the benefit only comes when you get up to 16," Vinall says. "I don`t think it`s really there today, but you can expect it."

Local carriers also should keep in mind two additional technical issues when evaluating the technology. First, say Vinall and Lynch, multivendor equipment compatibility can be problematic. "Today, the fiber-optic routers have to have the same equipment on both ends, because they don`t really have a universal standard for interconnecting disparate equipment today--like if you wanted to interconnect a Nortel and Fujitsu transponder at the optical level," Vinall asserts.

"And then there`s the issue of if you actually start, for example, interconnecting via wavelengths, where we just hand off a wavelength to a particular customer like a carrier or a customer, you have some issues there in terms of who owns the transponders," adds Lynch. "What wavelengths do you interface with, how do you ensure that particular customer is operating at that particular wavelength and not interfering with others on your system? There are a number of different issues that are going to have to be resolved."

These questions remain difficult to answer because of the other technical shortcoming--the lack of robust network-management capabilities. Again, a multivendor environment plays havoc with current network-management systems. "There`s a ton of development that has to be done in that area. It is not ready for prime time yet," Vinall concludes.

Points to ponder

With these pros and cons in mind, Vinall says that local-exchange carriers should consider three factors when evaluating DWDM in the near term. The first and most important of these is whether current bandwidth demands are large enough to make DWDM an immediate necessity. "People`s eyes are bigger than their stomachs right now," says Vinall. Therefore, a clear-eyed evaluation of current bandwidth demand is essential, as is the question of whether this demand can be met in other ways. Vinall doesn`t recommend upgrading that equipment with DWDM technology merely out of a reflexive desire to "futureproof" the network.

"The technology is being changed and deployed so fast," he asserts, "that if you don`t have the demand today for these services, sticking it out there because you think that hopefully 18 months from now you`re going to have a demand, you`re probably going to end up stranding capital. Because 18 months from now, you`re going to wish you had taken [advantage of] a new technology that no one`s actually deployed yet that will probably be more effective for whatever your market is going to be."

So how much bandwidth demand is enough? While some vendors would argue that OC-12 (622-Mbit/sec) or even OC-3 (155-Mbit/sec) traffic will support the expense of DWDM, Vinall suggests that OC-48 represents the current threshold. Lynch agrees: "Whether or not it`s reached OC-12, I guess is questionable. You`ll always get your best economics at OC-48 versus OC-12. That`s because you`re spreading it out over a larger number of bits."

In line with the notion of waiting for product evolution to intersect with an individual carrier`s current needs, Vinall expects that the development of less expensive metropolitan equipment will lower the bandwidth threshold in the near future. "It`s definitely dropped from OC-192s (10 Gbits/sec) to OC-48s (2.5 Gbits/sec). So I think because of that, you can pretty much anticipate that it`s going to go down to OC-12s," he says.

However, Internet service providers (ISPs) and other data-centric service providers who find DWDM appealing because optical internetworking may save them the cost of deploying SONET gear may prove a special case. "A lot of the ISPs, for example, don`t want to incur the additional overhead of putting [data traffic] onto a SONET system," says Lynch. "So in many cases, they want to just go straight on a transponder and into a dense wavelength-division multiplexed type of infrastructure. And if that is the situation, then you`ll see a lot more OC-3s and OC-12s deployed in a DWDM-type environment."

That said, the rapid evolution of technology could bite these carriers as well. With data switch manufacturers such as Cisco, Ascend, Jupiter, and Argon Networks announcing equipment with optical interfaces, adding these systems to the network in the future could strand transponders, says Lynch.

Once a carrier determines that the current bandwidth demand requires DWDM (and not just faster SONET gear), the next factor is ensuring that the equipment will operate with existing hardware, both in the host network and in the networks with which the DWDM network will be expected to connect. Vendors should be able to demonstrate interoperability with the existing network hardware before you sign on the dotted line, says Vinall.

Finally, carriers will need a detailed network plan to determine exactly how much equipment will be necessary--and whe ther that total expense doesn`t eclipse those of other options. Again, a realistic forecast of user demand is essential.

"When you put it on the network and you think about the network, you always tend to think about the ideal customer," cautions Vinall. "And the ideal customer buys from you at each span and at each node. But of course it never really works that way in real life. In real life, people buy a little bit here and a little bit there. That is what really could screw up that situation and cause the need for additional amplifiers. And I think that`s something that they would have to plan for when they plan out their metro network."

ISPs and other data-centric service providers who wish to take advantage of the optical internetworking aspects of DWDM also will have to plan their networks carefully. "Once you do have the datacom equipment, if the carrier is trying to obviate the need for SONET gear but you still want some sort of an add/drop architecture, you`ll need to do that add/drop via wavelengths as opposed to bringing it back up through the router. You can just keep it in the optical layer. It`s a lot more efficient," advises Lynch. That said, four wavelengths per site had better be enough, given the limitations of current equipment.

Protection switching also can be an issue without SONET equipment, Lynch says. SONET systems typically provide protection switching in 50 to 100 msec. ATM switches offer switching speeds in the hundreds of milliseconds, while router speeds can be an order of magnitude slower. A significant amount of data can be lost if protection switching isn`t addressed adequately, she says.

Thus, the rose of DWDM in metropolitan area networks bears its share of thorns, and local exchange carriers will have to be prudent gardeners as they consider the technology for their applications. Such carriers will have to weigh expense, present and future bandwidth demand, and network design intricacies against the known pros and cons of adding new fiber or beefing up SONET resources.£

More in Network Design