BOSTON (06/12/2000) - Have you looked at the price of Gigabit Ethernet equipment lately? It's dropping, and in a few months you'll see the floor fall out. At the recent NetWorld+Interop 2000, Broadcom Corp., a manufacturer of technology that goes into gigabit boxes, treated me to a technology demonstration. Broadcom makes the transmitter/receiver (transceiver) combinations that drive gigabit over fiber and the chips that make switching at that speed possible.
Optical multimode transceivers are due to drop to about half their current price as manufacturing economies of scale kick in, coupled with faster and more powerful LEDs. The copper wire side will fall as quickly or even more so as more organizations embrace gigabit. The only place we won't see startling price decreases will be single-mode fiber transceivers, which use lasers rather than LEDs.
There are two factors driving this connectivity explosion. The first one is simple: We need the bandwidth. There are more computers on larger networks with larger data sets everywhere you look. Gigabit network backbones are a necessity these days, as are gigabit links to selected servers and routers. The second accelerating factor is an emerging need to provide ultrafast access to the access router, if not the desktop. More and more eggs are going into the Ethernet basket, with voice over IP, multicast video and applications all vying for seats on the packet train.
This differs significantly from just a few years ago when the industry made the transition from Ethernet to Fast Ethernet. That process was accomplished in a much more leisurely manner, simply because there wasn't any real need for faster networks. A few organizations, mainly users of large Geographical Information Systems and shared Computer Aided Drafting networks, had the need for speed, but the rest of us really didn't care.
With the Internet explosion two years later, network bandwidth demands skyrocketed. Fast Ethernet was already cheap; cheap switches and adapters followed closely behind. What's surprising is how fast this process is taking place. If gigabit followed the same path as Fast Ethernet, we wouldn't expect it to become a commodity-level product for at least another year or so.
Instead, by the end of this summer you can expect gigabit to cost just slightly more than Fast Ethernet.
It's going to be economically feasible to run gigabit all the way to the desktop using the Category 5 copper wire you're now using for Fast Ethernet.
This is a wonderful opportunity and a potential problem. If we're going to run gigabit to the desktop, how do we construct a backbone that will handle it?
Broadcom has an answer in the form of 10G-bps Ethernet, an optically switched format that's already reached draft status with the 10Base-T Study Group (802.3ae).
At this rate, five years from now our networks will have petabit backbones with terabit drops; zottabit and yettabit networks won't be too far off. In case you're wondering, these are all technical terms that mean, "one followed by way too many zeros."
Shapiro is district technology coordinator for Kingsport City Schools in Tennessee. He can be reached at firstname.lastname@example.org.