If there's one thing that can be agreed on in computing it's that there's never one thing that can be agreed on.
Whether it's operating systems, bus design or CPU architecture, there is almost always at least one alternative available to make user purchasing decisions that much harder.
Networking is probably one of the areas worst affected. There are differing standards for everything from transmission protocols to cable design. Token ring has long competed with standard Ethernet frame format in the LAN market. Hewlett-Packard muddied the waters with its VG AnyLAN standard.
In the WAN market frame relay has competed with FDDI and Asynchronous Transfer Mode. While the first two of these technologies may have happily confined themselves to the network backbone, the ability of ATM to scale to the LAN is drawing it into competition with yet another emergent technology - 100Gbit/sec Ethernet.
But it is the emergence of Gbit Ethernet that poses the greatest challenge to ATM. Although the Institute of Electrical and Electronics Engineers (IEEE) is still probably a year away from determining the standards for Gbit Ethernet, a number of vendors are already making noises about this emergent technology. However, just as many are already making noises over ATM, and still more have lined up behind both camps.
ATM has a decided advantage over Gbit Ethernet due to the fact that it is already here. However, it remains for the moment prohibitively expensive for many organisations. Currently there are a comparatively small number of ATM installations in Australia, many of which are confined to classic "early adopter" sites such as universities and research centres. According to ATM vendor Fore Systems, ATM has received much broader acceptance in the US and Europe, with Australia lagging several months behind.
Fore Systems is a principal proponent of ATM technology. This is not surprising, considering the company was founded by four former ATM researchers, and its product line is comprised almost exclusively of ATM solutions.
Managing director Steve McRae feels it is too early to determine how much of a threat Gbit Ethernet will be to ATM. "At the moment Gbit Ethernet is still in the research and development stage, so there is a long path to go down before the real uses of that technology are clearly understood - whether it's viable and what the price points are going to be."
Apart from Fore Systems, IBM has also aligned itself exclusively with ATM through its alliance with ATM-developer Cascade Corporation. But the list of companies supporting the development of a Gbit Ethernet standard is formidable.
A so-called Gbit Alliance has been formed, with founding members including such notables as Cisco Systems, 3Com, Bay Networks, Intel, UB Networks and Sun Microsystems. Since then the number has swelled to forty, including Digital Equipment Corporation, Cabletron, Hewlett-Packard and a host of others.
A number of these alliance members, however, feel that both technologies will find a place in the marketplace, such as 3Com Corporation. Marketing manager at 3Com ANZA Andrew Hurt says 3Com already has working prototypes of a Gbit network running in lab conditions. "We definitely see it as being a co-existing technology to Gbit Ethernet. We don't see it as competing per se. It gets quite interesting when you look at the analogy of FDDI versus ATM; quite a lot of our customers are going with both at the same time. There's definitely advantages in both circumstances."
Cisco Systems is also pursuing both technologies, believing they can be combined to create effective corporate intranets.
IBM is siding itself with the ATM camp. Marketing manager for networking systems at IBM Australia Paul Kangro believes a number of questions need to be answered about how far Ethernet can be stretched.
"When Xerox proposed the original Ethernet it was clear that speeds of 1Gbit/sec were not on the wish list. With Ethernet you trade speed for distance because of CSMA/CD (Carrier Sense Multiple Access with Collision Detection), so for a Gbit Ethernet environment you could only have around 25 metres between the nodes - unless you changed the minimum frame size of 64 bytes - which means it starts to look less than Ethernet."
According to Kangro, because Ethernet uses CSMA/CD standard, the average maximum throughput is around 40 per cent utilisation before things start to slow down. "This means a maximum useable Ethernet of 400Mbit/sec - a lot less than ATM at 622Mbit/sec."
Indeed, there are many questions that still need to be settled about Gbit Ethernet. According to Fore's McRae, doubts remain about the ability of Gbit Ethernet to handle information forms such as voice and video, currently a strength of ATM. "It's purely targeted at the data side of the marketplace," said McRae.
According to Kangro, the ability of ATM to handle shared media on LANs makes it the superior choice. "ATM moves away from frame-based, shared media, multiprotocol networks towards a cell-based, switched environment that allows the LAN and WAN to work seamlessly."
However, Gbit Ethernet does have the advantage of requiring no data translation to carry information from conventional Ethernet networks - something required by ATM.
With the IEEE 802.3z study group still some time away from defining a standard, many of these questions will have to remain unanswered, leaving both vendors and users to guess as to how they will be positioned.
While not certain about Gbit Ethernet's position in the LAN, market development manager for network products HP Lindsay Lyon sees it as a clear competitor to ATM in the network backbone. "People are going to ask why would they go with 1Gbit/sec Ethernet if they can go ATM to the backbone? Well, you might not want ATM to the backbone, because you've got a start doing translational bridging if you put ATM into your backbone. Why would I even bring ATM into my backbone if I can get a Gbit/sec Ethernet and I have to do nothing - all the bridging's the same, the speed matching's the same, and there's no emulation. If Ethernet costs me say $2,000 a connection, and ATM costs me $8,000 a connection, and it will be at 655Mbit/sec not 1Gbit/sec, why would I do that?
"So I think Gbit Ethernet absolutely does threaten ATM in the backbone. I think that companies will feel a lot more confident putting in a faster Ethernet. ATM at 655Mbit/sec in the backbone is a very expensive thing to do."
Time, as always, will tell.