As the Special Reports theme this month illustrates, fiber-to-the-desktop remains a near-term goal for many in the optical communications community, particularly vendors. Technology developers have made significant progress in decreasing the cost of optical LANs. Small-form-factor (SFF) connectors and transceivers put optical networks on a par with copper as far as port density is concerned. The 100Base-SX standard opens a smooth migration from basic Ethernet to the multigigabit speeds of tomorrow. Centralized cabling architectures further streamline costs. Meanwhile, the bandwidth demands placed on LANs have begun to approach the level that many assert puts fiber squarely in play.
Things should look rosy, no? I checked this assumption a few months ago as part of an article I wrote for Lightwave's sister publication, Integrated Communications Design. Alas, the near-term prognosis for fiber in the horizontal doesn't look good.
Let's start with the SFF initiative. These connectors have yet to achieve significant popularity among network managers and designers. "No, they're not jumping on board rapidly," admits David Szczepanik, product marketing director at Sun Conversion Technologies (Quakertown, PA), a member of the TIA's Fiber Optic LAN Section. "It takes a pretty good sell job to get them to migrate to small form factor." Szczepanik cites the desire by many network managers to stick with the SC and ST connectors already in their networks as a major stumbling block to adoption of SFF technology. "There's still some questions in their minds about the small-form-factor standard. It is a little bit higher-priced. So if the wheel ain't broke, why fix it-they're going to stick with [legacy connectors]."
From the view of equipment manufacturers, a smaller connector alone doesn't guarantee parity with copper approaches. "It will lower the cost of fiber on a per-port basis because you can have more room on your front panel for more ports," says Bipin Mistry, technical marketing manager at 3Com (Santa Clara, CA), discussing SFF transceivers. "But the other thing you have to be aware of is, the more fiber ports you have on your front panel, the more heat you have to dissipate and the more power you require to drive those particular PHYs. There are still limitations on the number of ports you can actually physically put on a switch or a fixed-configuration switch or even a card for a chassis based on the dissipation and the power requirements for that particular box or card."
Even the jump in bandwidth requirements may not provide the push toward fiber some expected. According to Bruce Tolley, manager of emerging technologies and a self-described "1000Base-T evangelist" in the Gigabit Systems Business Unit of Cisco Systems (San Jose, CA), significant upgrades of copper cabling won't be necessary, as bandwidth demand increases to the gigabit-per-second range. "We believe that most customers will be able to run gigabit on copper if they have a cabling plant up to the 1995 Category 5 specification or a Category 5E specification," he says. Mistral agrees, saying that most of 3Com's customers installing new cable have chosen Category 5E copper for high-bandwidth applications in the horizontal.The daunting news for optical technology boosters is that if these sources at Cisco and 3Com represent the general wisdom among LAN equipment suppliers, fiber-to-the-desk hasn't won the minds of the hardware manufacturers whose opinions help shape the strategies of their network-manager customers. Until the LAN equipment companies buy into fiber, optical technology will remain a backbone solution.