Methodically visionary: The industry offers two different schools of thought on the best way to grow network bandwidth: through oceans of Sonet or towers of WDM. Carriers are relying on the best of both

One child builds a castle of sand, using a grand sense of vision to etch towers, battlements and flying buttresses from a mountain of crystals. A grand hall for a tiny king and all his subjects, a moat to protect the kingdom from would-be enemies.

Another child builds a house of blocks, using the sense of methodical plotting to create wall upon wall as logic dictates, room upon room as the miniature family of his mind grows. A garage for one car-no, make that two.

Different philosophies, but both model worlds are correct. The kingdom may grow, but the castle will always have room for more people. The family may grow, but just add another room when the new baby arrives.

Too bad theories on network expansion can't be reduced to such simple thinking. Or can they?

Some carriers are castle builders. They sculpt huge Sonet networks with grand bandwidth because they know the current nature of bandwidth growth puts Moore's Law to shame. They want to operate in all kinds of businesses, so they build networks that will support everything-including the traffic of other carriers as well as their own.

A castle builder's plans might have three phases: OC-12 (622 Mb/s), OC-48 (2.4 Gb/s), OC-192 (10 Gb/s). Growth is as simple as taking a big leap forward.

Other carriers build with blocks. They are obsessed with efficiencies in cost and labor. Why pay for capacity you are not ready to use? Let real need dictate expansion. Let growth be incremental rather than ahead of its time.

Their building blocks are channels created through wave division multiplexing (WDM). Growth is as methodical as a single step forward.

Both theories on growth pursue the same end: reliable, flexible networking in a world where bandwidth demand seems to know no end, but where network operations costs are a constant concern.

There has long been a question of which way is best, and advantages and disadvantages to both have materialized. Sonet infrastructure vendors and WDM developers have done their best to garner carrier notice. Ultimately, however, preference could be a matter of personal need.

The long and short of it WDM started as a murmur, a promising interim solution for satisfying surprising bandwidth demand. A few years after its auspicious debut, however, it is a full-fledged business. Several long-distance carriers have deployed the technology, especially in point-to-point configurations, to satisfy the needs of major clients.

In recent months, the WDM business has entered its second phase. Vendors are targeting lower-cost designs toward the local exchange market, particularly metropolitan areas where carriers are mulling how to beef up the OC-12 or OC-48 Sonet architectures they never thought they would have to beef up.

This need to improve network muscle coincides with the most forceful conservatism our already thrifty industry has ever seen.

Local exchange carriers, especially the Bell companies and the smallest debt-ridden new carriers, are not just looking for inexpensive bandwidth enhancement. They also want growth they can control and technology that can create incrementally more bandwidth as unpredictable demand sees fit to show itself.

"All carriers have to make the bandwidth decision. For example, when do I go from OC-12 to OC-48?" says Jesus Leon, vice president of access products at Ciena Corp., currently No. 1 on the WDM pop charts. "It's easier to add waves than it is to add new Sonet equipment."

The argument for WDM may get even more compelling as carriers consider the wholesale jump from OC-48 to OC-192. Aside from the largest interexchange carriers, most carriers don't have an apparent need to fill that kind of capacity.

"That's a bleeding edge improvement that requires a disruption of service," says Jim True, vice president of marketing at Osicom, another WDM player.

WDM has become an increasingly popular method of bandwidth growth over the past year or so, and people within and outside of the industry have confirmed its importance to the future of networking.

Yet the technology has not been truly tested in the one arena where philosophies of incremental growth may make the most sense. Many vendors have jumped on the bandwagon aiming to bring WDM to local point-to-point applications or metropolitan area rings, but only a few products are generally available. Although many LECs have dabbled with WDM, most haven't yet revealed plans for how extensively they will use the technology.

Lower prices for LEC-targeted WDM systems will win carriers over. However, a persistent management issue may keep WDM's role in local networks limited for now.

Many short-haul WDM systems-the majority of which started as long-haul systems for IXC networks-were not designed with element management system interfaces in mind.

"Carriers are aware of these management issues, which increase network complexity the more WDM you put in," says Brian McFadden, assistant vice president of brand management for Sonet networks at Northern Telecom.

Nortel, like most other Sonet infrastructure vendors, prescribes WDM usage in limited doses where capacity needs are obvious. However, the company also has sold more ultra high-capacity OC-192 Sonet configurations than other vendors and maintains that Sonet bit rate expansion should be the basis for network capacity growth.

"With WDM, when you disassociate a wavelength from a fiber, judging the quality of service is difficult," says McFadden.

Although multivendor Sonet networks have introduced their own complexities into today's networks, in the last year or so, vendors have become much more progressive about helping carriers build easy-to-manage multivendor Sonet environments. This has been accomplished through a combination of a new cooperative spirit among vendors and increasing integration between new, robust and flexible Sonet network element management systems and comprehensive network management operations.

That, in turn, makes upgrades from OC-12 to OC-48 and OC-48 to OC-192 appear more manageable.

"Sonet equipment senses that service has been interrupted. That kind of intelligence is not currently built into WDM," says Marty Kaplan, chief technology officer at Sprint.

Sonet infrastructure vendors also are promoting Sonet bandwidth upgrades through card installments rather than equipment swap-outs.

In addition, Sonet bit rate expansion may be an altogether easier consideration for carrier operations' depleted staffs. Wholesale upgrades guarantee bandwidth through uncertain traffic growth patterns, and for new Sonet carriers dealing with initial growth spurts, going from OC-3 to OC-12 may be simpler and more cost-effective than adding wave upon wave after every customer base fluctuation.

Ciena's Leon agrees that WDM addresses certain stages of network growth, acknowledging that administering WDM to an OC-3 configuration, rather than moving straight to OC-12, might not help a carrier realize measurable cost efficiencies.

However, WDM is not being positioned as a network buildout option. It's strictly a method of traffic accommodation that actually relies heavily on existing network facilities. It has been proved, through carrier acceptance, that it is not only worthy of the challenge but also can be the tool of logic and cost efficiency that carriers need.

"WDM is great for incremental growth, and lighting up new windows is simple," says Kaplan.

The marriage of methods

So far, while investment in WDM increases, interest in OC-192 Sonet systems seems to be exceeding past expectations. Nortel's customers include MCI, Qwest Communications, GTE Internetworking and Canada's Fonorola. Other carriers such as Sprint have been testing OC-192 in lab environments.

Meanwhile, vendors have even talked of pushing that envelope. "You can increase Sonet base rate frequencies to multiples of OC-192," says Larry Seese, executive vice president of network and operations planning at Qwest.

"OC-768 could be next. Four times the current rate has been the logical step for Sonet. But, it may take a few years as we figure out how to manage that growth," says Nortel's McFadden.

It is unclear how soon carriers such as Qwest will require such bandwidth, although Seese appears anxious for the chance to try it.

While such carriers are hungry for OC-192 systems, it can be said that such massive bandwidth appeals to a somewhat narrow carrier profile. Nortel's OC-192 customers have one thing in common: They are broad-even national-aggregators of long-haul traffic looking to accommodate rapid growth in Internet-related data networking. Also, some of them, most notably Qwest, fancy themselves as operators of transport networks that will cater particularly to non-facilities based carriers, such as Internet service providers or resellers. This means that their networks must support the weight of several carriers.

"OC-192 is strictly the highest-capacity, lowest-cost position. As data grows by 100%, 200% or 300%-take your pick-bandwidth is your only limitation. Having a high base rate frequency is fundamental to our strategy," says Seese.

For GTE Internetworking, extraordinary Internet bandwidth demands set the stage for OC-192. "Internet traffic is doubling every four to six months for us. OC-192 satisfies our immediate needs and long-term growth," says Steve Blumenthal, vice president of engineering at GTE Internetworking.

Still, WDM also has its place for such carriers, particularly in markets where bandwidth growth is most intense. "We can do up to eight-channel WDM [with OC-192] right now if we want," says Seese. "There are some deployment cases where we want an extra wave in there, or some cases where we might light up two windows."

MCI already is practicing this, having launched OC-192 with eight-channel WDM this month.

"When we have OC-192 ready for our network, we'll also do 16-channel WDM," adds Sprint's Kaplan.

However, while network traffic in local exchange networks also will grow rapidly, the short-haul, point-to-point nature of this traffic probably will not require OC-192 capacity any time soon. Most LECs with Sonet have OC-12 configurations, or OC-48 in some extreme cases. For them, WDM systems, if they meet cost and management requirements, will become increasingly attractive options.

"We're really talking about different models and a whole set of different issues," says Mathew Steinberg, senior analyst at Ryan, Hankin and Kent in South San Francisco. "IXCs were willing to take a risk on deploying WDM early because fiber exhaust was so bad for them. But as long-haul aggregators of traffic coming from many points, they need high bit-rate Sonet.

"For the local guys, they will want to look long and hard at future management support for WDM. They need to do more things remotely, such as configuration and monitoring, and need to have control over that," he says. Meanwhile, they still won't have much need for high-bandwidth traffic aggregation.

Although each world has its distinct needs, it is apparent that base rate Sonet upgrades and gradual WDM increases will be complementary in satisfying the requirements of both.

"We have to make WDM available to our customers as a matter of flexibility," says Nortel's McFadden. "I'm a WDM guy, too."

Even WDM developers, some of which have claimed that WDM will eventually displace Sonet base rate upgrades as the dominant method of network growth, can see the complementary reality.

"WDM doesn't displace Sonet at all," says Osicom's True. "It isn't competing with Sonet as a structure but is complementary to Sonet's growth. This combination is the ultimate low-cost approach."

Carriers agree. "WDM is for fiber replacement, not Sonet replacement," says Sprint's Kaplan.

Which way the combination leans will depend entirely on the needs of the carrier in question, but carriers already recognize the respective benefits of these technologies. Says Qwest's Seese: "Our Sonet base rate combined with our WDM capabilities gives us the best of both worlds."