After we revived her--by ruffling a few recent and very thick issues under her nose and assuring her that we did indeed see ourselves as a bit more strategic than our contributor's comments suggested--we continued our discussion of the imminent demise of Token Ring and the likely healthy future of Unix. She felt a little better as the conversation continued, but to this day that contributor still calls me an Ethernet bigot for my lack of willingness to consider Token Ring as a useful technology.

Like Bringing a Knife to a Gunfight And so, I am an Ethernet bigot. I remember reading early on about Token Ring's elegant media access algorithms and thinking about how clever it was compared to Ethernet's far simpler random retry algorithm. But when I actually had to write requisitions for equipment, I quickly became a convert to Ethernet. The market has shown that, like bringing a knife to a gunfight, cheap and effective blows away elegant and expensive every time. You'd think that the Token Ring faithful would see the handwriting on the wall and just give it up. But no, there are always those who will try to force a market to exist where none should.

Those who have bet the farm on Token Ring struggle to squeeze the last bit of revenue out of a technology at the expense of the best interests of their customers. High Speed Token Ring (HSTR), which runs at 100 Mbps, has recently reached "implementable draft standard" status, and a few companies, including Olicom and Madge Networks, which have a long and vested history with the technology, have announced HSTR products. But the price and selection of HSTR equipment alone should convince even the most stalwart Token Ring holdouts to switch to Ethernet.

Shortcut to a Standard HSTR's inventors understood that if they came up with a unique physical layer specification, the cost of their technology likely would be prohibitive and the time to market would be tragically long and drawn out. So, they've borrowed Fast Ethernet's physical layer and stuck the Token Ring media access control on top of it. This is a fairly common practice, but it means abandoning older Token Ring wiring. And the instant we start talking about rewiring, there isn't much incentive to stick with a technology like HSTR anyway, since it is clearly falling behind Ethernet both in terms of price and performance.

Still, there are those diehards who just can't walk away from the idea of Token Ring. If you're one of them, I have to wonder why you didn't buy FDDI. After all, it's token ring at 100 Mbps. The cold, hard reality is that network design has left the media-access arguments far behind. It's increasingly rare to find a network design in which end stations aren't on very small shared segments or on completely private segments. Ethernet is essentially just a fancy serial port and that's exactly what we need for today's highly segmented switched networks.

Token Ring Is Dead, Leave It So It's time for the HSTR crowd to abandon its parochial interests and serve its loyal customers by bringing them into the Ethernet world rather than by planting the notion that there is any reason to continue implementing Token Ring networks at any speed. Just let it die.

Send comments on this column to Art Wittmann at awittmann@nwc.com.