Raising the standard

I consider myself fortunate to have been involved in a number of standards committees during my career. Taking a new technological idea and creating a billion dollar market is something in which every engineer dreams of participating. To shape the future of the networking industry provides a great deal of satisfaction. This happened with Fast Ethernet and will surely happen in the future with Gigabit Ethernet.

The IEEE first got together in August 1993 to discuss a 100Mbps revision to the 802.3 standard. This was in response to two market factors. Firstly, users required a high-speed, backbone-to-workgroup networking standard that didn't have the same high product and media costs as FDDI. Secondly, Ethernet was in the process of dominating Token Ring as the networking technology of choice and so had a guaranteed future. Despite these factors, the discussions over the most appropriate technology was taking time; mostly due to debate between three competing technologies. Those competing for IEEE approval were the Thomas Conrad Networking System (TCNS), the 100VG-Anylan and the now ubiquitous 100Base-T.

TCNS was supported by Thomas Conrad, a start-up company later acquired by Compaq and 100VG-Anylan which was supported by AT&T and Hewlett-Packard.

However, both eventually switched their allegiance to 100Base-T. VG-Anylan was based on a technology called Demand Priority Access Method (DPAM).

This demand-response technology was arguably technically superior to 100Base-T but it was also a new technology and different from Ethernet. VG-Anylan did become an IEEE standard but as 802.12, not the popular 802.3. If it's not in 802.3, it's not Ethernet and that caused 100VG-Anylan to die a slow, but dignified death.

Opposing standards

The 100Base-T proposal drew the most support, with backing from Intel, 3Com, Synoptics (now Bay Networks), SMC and Sun Microsystems among others.

I was one of the Intel representatives and was involved in these initial confrontations between opposing standards.

The disastrous side effect of the vigorous debate between these competing 'factions' was that it delayed getting a workable standard agreed. In order to speed up the ratification process a number of the 100Base-T advocates formed the Fast Ethernet Alliance (FEA). Its objective was to produce a workable addition to the 802.3 standard that the IEEE would approve.

However, the stalemate did succeed in uniting the 100Base-T supporters (most of the industry) behind a common cause - speeding up Ethernet.

The Fast Ethernet Alliance was originally made up of personnel from seven vendors, representing over 80 per cent of the total networking market.

The common feeling within the alliance was that agreement of a standard should take priority above all else. Without a workable amendment to 802.3, we were all in danger of missing out on a huge potential market and simultaneously frustrating users who were crying out for more bandwidth.

Without the standard being agreed, there was no market. Competition was put aside as the other vendor representatives and I took a two-phase approach.

Phase one was to get the standard ratified. Phase two was to bring the products to market and start competing.

Lessons learnt

When the agreed Fast Ethernet technical specification was delivered to the IEEE, it was the first time since the original Ethernet standard that a commercial group of industry heavyweights had formed an alliance to develop a common standard. The key difference was that this made life easier for the IEEE, which by late 1994 not only had a technologically robust proposal but also knew it would be accepted by 80 per cent of the market without argument.

The 802.3 revision was finally ratified in June 1995, a lot faster than it would have been if not for the Fast Ethernet Alliance. And we certainly learnt lessons to be applied in the race to ratify Gigabit Ethernet.

When it became clear that the market required an even faster backbone speed, the Gigabit Ethernet Alliance was formed to use the experience gained from the Fast Ethernet Alliance. Using this experience, the time from concept to approved standard was cut from three to two years. Gigabit Ethernet (802.3z) was ratified as an official IEEE standard in June this year.

Unsurprisingly, the take-up of Gigabit Ethernet has been slow and steady.

Companies are starting to pilot Gigabit projects and evaluate the technology available from vendors like Intel. But I can see the market exploding next year, when the hard work of the standards committees and the alliance will all be worthwhile. Already network managers have a number of interoperable Gigabit Ethernet products to choose from, where you can basically plug one end of a cable into a Cisco switch and the other into an Intel adaptor and it works. That's a well-designed standard in action.

The natural progression should be Gigabit Ethernet moving closer and closer to the desktop. While the applications that require this bandwidth may not be here now, who knows what the future holds?