From the George Gilder thread. This article thinks SONET will be around for some time yet.
The Advent of WDM and the All-Optical Network: A Reality Check
Some forecasters in the industry seem to be getting carried away with the influence of WDM in the network and visions of the all-optical network.
July 1998
------------------------------------------------------------------------
As the telecom industry has entered the nascent stage of the market hype cycle for the all-optical network, it is useful to remember history in order to keep expectations in perspective. In the industry, there are very few examples of new technologies meeting original expectations of market timing. In fact, many predictions have been way off the mark, especially in relation to the degree of market penetration. ISDN has been excruciatingly slow in comparison to the initial discussions of the technology many years ago. SMDS fell way short of expectations. SONET was a long time in getting to its current dominant position in the public network.
Rare situations that occurred have led to big jumps in particular market segments. Soon after the first divestiture of AT&T in 1984, sales of fiber optic cable and electronics soared as MCI and Sprint raced to build up their long-haul networks to compete with the dominant interexchange carrier (IXC). Using this extraordinary event as an example, in the late 1980s, some analysts were mistakenly predicting massive deployment of fiber to the curb or home (FTTC/FTTH) during the 1990s. It is now apparent that we may be well past the year 2000 before fiber in the last mile becomes a reality.
Similarly, wavelength division multiplexing (WDM) has taken off in the last couple of years, but largely because of an irregular set of circumstances. Large IXCs, including AT&T, WorldCom, and Sprint, did not have a time division multiplexing alternative. (OC-192 took a while to become available and it was cost-prohibitive to place additional fiber in the long-haul networks.)
Consequently, WDM became a major solution for dealing with capacity constraints. In recent months, there has been a tremendous amount of hype that WDM sales will explode in the metropolitan markets in a similar fashion. Although industry forecasters will probably not be as dramatically off with FTTC/FTTH prognostications as they were in the late 1980s, some seem to be getting carried away with WDM becoming ubiquitous quickly and with visions of the all-optical network becoming a reality relatively soon.
Inflated Estimates?
Some of this exuberance is undoubtedly spurred on by inflated WDM size estimates partially caused by double accounting practices (totaling internal component sales a second time as part of system sales) and perhaps by including non-WDM gear in the total sales. It is possible that, in 1997, the U.S. WDM system market did not even exceed $800 million, close to half the amount of estimates that have been floated in the market. It's also possible that this market may never exceed $1.2 billion in the foreseeable future because of significant price declines, regulatory matters, and alternative technologies. Incredibly, other analysts are forecasting the market to reach more than three times this projection.
In general, change in telecom is extremely slow. Despite a lot of hype about a new industry environment through deregulation (which is really re-regulation), it remains a conservative marketplace.
While WDM sales will continue to increase in a more modest fashion, there are no self-evident pockets of opportunity that will propel them anywhere near the growth that has been experienced over the last two years. Although prices continue to decline on WDM gear, it is not clear that widespread deployment will occur in the metropolitan environment in the very short term, except for cases where there are extreme network facility exhaust scenarios--in densely populated, urban areas where it is difficult to dig trenches to put in more fibers or even to build new conduit.
Bell Atlantic, which appears to have a disproportionate number of these conduit or hard structure exhaust situations compared to the rest of the regional holding companies (RHCs), may be the only local carrier to deploy WDM in a big way in 1998. Some government agencies actually would require that streets be rebuilt from curb to curb instead of just digging trenches and repairing the areas. The cost of WDM is a lot more attractive than digging up the streets.
The other local exchange carriers (LECs) and competitive LECs (CLECs) do not seem to be in as much of a crisis situation. Moreover, there are alternative solutions to WDM that are viable. In many cases, adding more optical fiber looks attractive. OC-192 systems will mature and offer the full ring capability as well as the granularity that local carriers desire. Furthermore, traffic can be moved around a ring to achieve more capacity. So there will be room for a variety of solutions and the choices may be decided on a case-by-case basis.
With WDM taking a longer time to reach critical mass in the local environment compared to the interexchange market, the evolution to networking at the optical layer will be gradual. Given the more mesh-like networks and shorter distances of the LECs and CLECs, they will eventually be the biggest buyers of optical add/drop multiplexers (OADMs). Major customers such as financial institutions are getting increasingly specific about where they want to transport traffic. Conversely, the long-distance carriers tend to be point-to-point with a great deal of aggregation and fewer routes.
Programmable (software-configurable) OADMs will probably not be deployed in volume for nearly two years. At present, there are only fixed varieties (manual patch) that do not allow for flexibility in dropping or adding particular wavelengths. We at Trans-Formation project that the U.S. OADM market will grow to about $420 million in 2002.
The next stage in the evolution to optical networking will be cross-connecting and switching wavelengths. These products are not likely to be deployed for at least five years and the price of these complex, high-capacity systems will be extremely high.
The industry is very early in the gestation period of optical cross-connect (OCX) deployment. A switching fabric needs to be developed that will allow for true scaleability. Ultimately, hundreds of ports will be required.
What About SONET?
The most important factor that will delay wavelength networking may be the lack of both performance monitoring and network management at the optical layer. There is a very well-defined management scheme with SONET. The problem of tracking and maintaining wavelengths across a complex network will be a significant hurdle. A mechanism needs to be devised for failure detection with optical switches that will allow communication back to a central processor to establish a switch-over to protection/restoration capacity.
Predictions that the optical layer will replace SONET are farfetched. The carriers are not going to throw out their tremendous investments in SONET operations systems overnight. Just as there are several different network elements to deal with capacity constraints, there will not be an exclusive layer in the network to transport traffic. The optical and perhaps the ATM layers will complement the existing SONET layer.
Regarding future considerations for wavelengths and networking, tunable lasers are going to be increasingly desired for WDM systems, especially relating to the next-generation gear. Major carriers that deploy large quantities of this equipment are going to want to limit the amount of investment in spares. It can be a logistics pain in large IXC networks to even deal with 10 different lasers, let alone stock for 40 wavelengths. Four to six lasers that can tune across all or half of the band will be needed. However, it is uncertain how long it will take for these tunable lasers to become commercially available.
The need for WDM at the high end of the market will not go away because the next-generation TDM systems may not be available for a very long time. A whole new generation of opto-electronics will need to be developed before the arrival of OC-768. Moreover, there are doubts whether such a system can work practically over the existing fiber in the network. However, 16 channels working on an OC-192 system may be enough capacity in almost any large network for quite a long time, except for niche locations where there is limited fiber.
Optical TDM
Another solution that can theoretically impact the optical networking vision is an "optical TDM." Suppliers have started to tell carriers to expect such a device in three to five years. (Given the track record of suppliers with availability dates, it is probably safer to assume seven to 10 years.) This next-generation simplification will multiplex multiple streams into one digital system--apparently eliminating the need for windows. There are always efficiencies to generating one signal form rather than 40 signals. This technology has been researched in laboratories for years. Test beds exist but not quite to the system speed of today.
While an optical TDM can theoretically make the need for a window hierarchy disappear, by the time it becomes available, there will be a huge installed base of wavelengths in the network that will still need to be managed. Again, the lesson from history is that the public network does not tend to remove equipment in its network in a hurry.
Mark Lutkowitz is president of Trans-Formation, Inc., an independent research firm in Birmingham, Ala., specializing in transmission equipment markets. He can be reached by telephone at (205) 970-2250 or via e-mail at trans@wwisp.com. |
