Gigabit Ethernet and ATM race along the backbone

Gigabit Ethernet and ATM race along the backbone

George Kotelly

Executive Editor

According to communications analysts, manufacturers, technologists, and users, Gigabit Ethernet and Asynchronous Transfer Mode (ATM) technologies are expected to compete vigorously for implementation in future, very-high-speed, fiber-optic backbone networks.

Network planners and managers are primarily looking for accelerated platforms to handle future multimedia and videoconferencing applications at the server and backbone segments. Furthermore, they also need solutions for their Internet access, e-mail, and database warehouse requirements.

Some industry experts feel that the two technologies will, instead, target a specific infrastructure--Gigabit Ethernet in local area networks (LANs) and ATM in wide area networks (WANs). In either implementation, key capabilities have to be established as to standards, interoperability, scalability, quality of service, and migration. Still other pundits predict that both technologies could work together with ATM in the backbone and Gigabit Ethernet in the local loop.

Whereas industry enthusiasm for ATM has diminished perceptibly (see Lightwave, June 1996, page 4), Gigabit Ethernet has spawned the establishment of more than 85 vendors in a technology support group called the Gigabit Ethernet Alliance. In addition, at a recent meeting, the Institute of Electrical and Electronics Engineers` 802.3¥Gigabit Ethernet task force modified the basic tenets for the technology and issued a draft specification. It also agreed to send the latest version of the specification to an ieee letter ballot this July, and to predict final standardization in March 1998.

Even though Gigabit Ethernet standards will not be available until next year, Dataquest, a San Jose, CA-based market research consultancy, forecasts that this technology will emerge as a leading section of the networking industry. It estimates that Gigabit Ethernet-related revenues will total $74 million in 1997 and $2.9 billion worldwide by the year 2000.

According to Trudy Barker, director and principal analyst in Dataquest`s telecommunications group, "Gigabit Ethernet will become the dominant backbone technology in the near-term planning horizon, barring any lurking killer application for ATM. Gigabit Ethernet will be used in data-intensive environments."

With a strong foundation in two well- established technologies--Ethernet and Fibre Channel--Gigabit Ethernet transmits data at an effective 1-gigabit-per-second rate while retaining Ethernet`s framing format and managed-object specifications. Industry studies indicate that network planners and managers would pay double the cost of existing equipment if they could gain a tenfold boost in speed.

Dataquest`s director and principal analyst of networking services Dan Miller claims that Gigabit Ethernet is simple, fast, and cheap, with a cost per node of about $1600, compared to nearly $4000 for today`s 622-Mbit/sec ATM. His company gives the market edge to Gigabit Ethernet because every major vendor has announced support and the standards are evolving quickly.

Some network planners and managers, however, question the need for higher data speeds for some time to come and do not welcome the expense of new equipment, technology, and training. Other changes involve the possible installation of fiber-optic and copper cables. Fiber cables are expected to be installed to the nodes with coaxial cable or copper twisted pair to the user. New network interface cards must be installed as well.

Furthermore, both ATM and Gigabit Ethernet technologies are mired in incomplete standards. Full-scale implementation of ATM technology has been bogged down by numerous specification changes, expensive equipment, and complicated deployment. Standards for Gigabit Ethernet are available but will not be fully sanctioned for more than a year. In addition, such fundamental issues as shared media, switching, congestion, and distance functions are still being resolved.

A larger barrier, though, is inertia. With more than a reported 60 million Ethernet nodes in use, many users are expected to wait out both ATM and Gigabit Ethernet implementations until the technologies completely prove their potential in the marketplace.

Based on the market`s delayed acceptance of ATM, the potential of Gigabit Ethernet`s awesome speeds at reasonable prices will evolve only when standard products can be interfaced and interoperated properly. q

More in Market Research