Despite this critical step forward for Token Ring users around the world, the technology is besieged by naysayers who refuse to grant this advance its due respect (ý la Art Wittmann in his column). It is all too easy for the luminaries of our industry to make a religious war of the Token-Ring-versus-Ethernet issue. However, Token Ring users don't care about religion; they want solutions that will meet their needs today and lay the groundwork for future growth.

Today's Token Ring customers have a wider array of upgrade options than ever before, including switched Token Ring, HSTR, ATM and Fast Ethernet. Each of these options plays a vital role in implementing the best path forward.

The real value of HSTR solutions is that they plug in seamlessly to a switched Token Ring network, providing an immediate, cost-effective performance upgrade where networks need it most: at the server and the backbone. The value proposition for Token Ring customers is clear: Install a sixfold bandwidth increase for less than $750 per connection (switch port plus NIC), without any changes to the surrounding network infrastructure (including cabling). HSTR is the perfect, even necessary, complement to Token Ring switching.

As enterprises increasingly employ Token Ring switching to alleviate congestion in the workgroup, greater aggregate bandwidth is pushed onto the backbone and to file servers, causing potential bottlenecks. HSTR alleviates these two specific choke points.

Today, there is an immense base of Token Ring customers (more than 22 million nodes, according to market research firm IDC) that stands to benefit significantly from HSTR. The widespread acceptance of Token Ring switching provides insight into how today's Token Ring managers view the future. Market research company the Dell'Oro Group reports a near doubling of the total worldwide market for Token Ring switching from 1997 to 1998, and forecasts continued strong growth through the year 2000. The growing adoption of Token Ring switching will drive the need for HSTR solutions.

The vote has been cast about the future of Token Ring by the people who matter most: the customers.

While some Token Ring proponents may dwell on the technical superiority of Token Ring to quiet the Ethernet fanfare, the essential reality facing many Token Ring customers is that a forklift upgrade to Ethernet will not work.

Having built their networks and business applications around the characteristics of Token Ring, such as large frame sizes and Source Route Bridging, a move to Ethernet would entail an overhaul of such magnitude that it simply could not be justified on the basis of performance or cost. For these enterprises, HSTR represents a welcome solution that will meet the need for bandwidth well into the foreseeable future.

Alternatively, Ethernet proponents wield the cost argument to advocate a wholesale changeover to Ethernet. However, when it comes to improving Token Ring network performance, comparing the cost of Ethernet gear to Token Ring is comparing apples to oranges. For new installations, Ethernet is clearly the less expensive choice. But for Token Ring users, ripping out Token Ring in favor of Ethernet entails significant hidden costs in both installation and reconfiguration that cannot be overlooked.

It is unjust--reckless even--for those in positions of influence, such as Art Wittmann, to counsel the abandonment of a technology that has served its users reliably for more than a decade and shows no signs of letting anyone down. Wittmann's advice for Token Ring vendors to "abandon their parochial interests" and let Token Ring die is like telling Macintosh developers to give it up because the PC is where it's at.

Worse, Wittmann makes it sound as if Token Ring vendors are doing their customers a disservice by pushing HSTR, while we have found the exact opposite to be true. Our Token Ring customers' response to HSTR has been overwhelmingly positive. Wittmann should talk to a few of them before he calls Token Ring down for the count.

Jýrgen Hýg is Olicom Corp.'s vice president for product marketing. He can be reached at jhog@olicom.com.

I know that many of you have both Token Ring and Ethernet in your shops, and that the Token Ring side is probably running out of steam. Token Ring's developer IBM knew this a few years ago, and was developing 25-Mbps ATM as a transition strategy. That's no longer a strategy, but the concept is telling. If you need more bandwidth in your Token Ring network, you now have choices: You can begin the process of removing Token Ring--most likely going to Ethernet--or you can use some technology to break up your Token Ring networks with switching to introduce more bandwidth. You can do that in a number of ways, using ATM, FDDI, Ethernet or, now, HSTR.

If you decide to upgrade the back end of your Token Ring network, you'll eventually want to connect it to the rest of your network. And herein lies the problem. You'll do that with a router, with which you can implement all the rules and access-control restrictions that you have between your Ethernet and Token Ring networks. Most of you will do that with Cisco Systems routers. But Cisco has no plans to develop HSTR technology and it does not belong to the HSTR alliance. Why? Because Cisco thinks the way to develop networks is with other technologies. Cisco is by no means an Ethernet bigot; it sells ATM, FDDI, and both 16- and 4-Mbps Token Ring. But, Cisco will tell you, the future of the desktop is brightest in Ethernet.

Even if you don't use Cisco routers, the picture is bleak. Bay Networks, for example, is a member of the HSTR alliance, but when I asked about HSTR development plans, no one at Bay was able to share anything with me. I specifically asked if there would be an HSTR interface for Bay routers, and I couldn't get a commitment that one would ever exist. Building HSTR networks is like building an island with no port--it might be a very nice island, but eventually you'll want a way to get back to the rest of world.

The HSTR vendors very carefully spin the advantages of the technology. They tell you it's only marginally more expensive than 16-Mbps Token Ring, and that's true. However, they don't compare the cost of HSTR to the cost of Ethernet, particularly when most of today's PCs have a 10/100 Ethernet port on the motherboard. They tell you HSTR will run on existing wiring as well as single and multimode fiber, and that's also true. But they don't tell you it's true because HSTR used the Fast Ethernet physical media specification--anywhere that one runs, the other does too.

In the short term--and I mean the very short term--HSTR might solve a problem or two and thus might be a fine tactical purchase. But please don't bet your network on it for the long run. There was indeed a technology war. Token Ring and Ethernet fought it out for the desktop, and by a roughly 10-to-1 margin, Token Ring lost. If you are in a position where you need to add significant throughput to your network, consider this to be the ideal time to start moving away from Token Ring. The innovation in the industry is happening elsewhere, and that's where you need to look.

Desktop and laptop computers have about a four-year useful life in most corporations. You need to be willing to rethink your networking strategy with that time frame in mind. With Bay, Cisco, Compaq Computer, Intel and 3Com all squarely in the Ethernet space, your Token Ring network won't make sense down the road.

Rather than thinking about quick fixes with HSTR, think about a long-term strategy that's more closely aligned with the vast majority of the networking industry. Forklift upgrades aren't necessary and never have been, but a good evolutionary strategy is essential; HSTR just isn't a part of it.

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