TCP/IP will overtake ATM soon

According to Ovum, broadband networking will use a wide variety of protocols influenced by the following factors: Customer requirements and the needs of underlying applications LAN developments and the adoption of Ethernet enhancements; adoption of ATM within the telcos wide area networks and adoption of TCP/IP based applications and voice over IP. The interaction of these factors will determine to which extent ATM and IP will be used.

For many years, the orthodox view of the development of broadband networking has been that it would use ATM (asynchronous transfer mode) transmission and switching. The definition of the broadband ISDN has been ATM on an SDH (synchronous digital hierarchy) infrastructure. The ITU-T has been developing ATM standards for the B-ISDN for nearly 10 years. Telcos and equipment suppliers have co-operated in the development of these standards. They have developed a vision of the future accepted by regulators and end-users alike. The path towards this future has been well marked - although exactly when it would be realised has remained unclear.

The rise of the Internet and enhancements to the TCP/IP protocol suite to make it better able to cope with multimedia, broadband and real time communications has challenged that orthodoxy. The Internet has grown at rates which are terrifying for a traditional telco - 10% - 15% per month. The Internet has a number of problems; it is slow, it is insecure, it can't cope very well with real time applications and has no end to end management. But all of these problems are being addressed. There are proposals for improving real time applications (RTP), for reserving bandwidth (RSVP) and improving security (IPSEC).

According to Ovum customers requirement for broadband is being driven by the growth of traffic and the need to reduce latency rather than by widespread adoption of "broadband applications". The key drivers for broadband are increasing use of client - server architectures, the growth in e-mail and the increasing size of files being transferred across the wide area. Voice traffic remains a major component of wide area broadband. Inherently broadband applications such as real time video or CAD/CAM are contributing a relatively small amount to the requirement for broadband access capacity. E-mail is a major factor because of the attachment of large files - spreadsheets, presentations, voice messages, etc. The Worldwide Web is also important because users wish to download complex screens quickly.

Non-voice is overtaking voice as the prime driver for wide area broadband. Data has been growing faster than voice for many years, but until recently, the total amount of voice traffic exceeded non-voice. They are now roughly equal and in a few years, non-voice broadband traffic will be twice as great as voice. Within 5 years, large corporate customers will treat voice as one application on their broadband networks - albeit an important one.

Gigabit Ethernet and Ethernet switching are reducing the need for ATM in the campus backbone. Gigabit Ethernet allows LANs to operate at 1,000 Mbit/s and switches are available which allow switching at wire speed. As a result, there is less need for 625 Mbit/s ATM systems in the campus backbone. Gigabit Ethernet is also displacing FDDI in the campus backbone. Although Gigabit Ethernet requires optical fibres instead of wire, it uses the same frame structure as traditional Ethernet, once again leaving users feeling comfortable that it is a low risk option./P>

Thus, there will be no strong driver for adopting wide area ATM from developments in the LAN. If ATM were being adopted at the desktop or in the campus backbone it would be natural to use ATM wide area services. In the absence of ATM in the LAN, wide area ATM services will stand or fall on the intrinsic advantages they offer to users. Ovum concludes that IP will be the dominant protocol for broadband wide area networking by 2002, concludes Ovum. Use of both ATM and TCP/IP will grow rapidly but by 2002 usage of TCP/IP will exceed usage of other protocols. The IP market will grow rapidly at more than 70% a year between 1998 and 2002. The ATM market will also grow rapidly but will be less significant than TCP/IP.