The New Universal Network Solution, or the Next ISDN Pipe Dream?

Asynchronous Transfer Mode (ATM) is a broadband, low delay, packet-like switching and multiplexing technique that segments voice, data, multimedia and video into fixed size cells for transmission. It has moved from its early hype as the be-all, end-all network solution to a practical approach for merging many applications while providing quality performance.

The course of ATM deployment has defied the expectations of most market participants. Originally, ATM was felt to be the optimum solution for integrating private networks and providing desktop-to-desktop connectivity around the world. As a result, it was assumed that ATM would move rapidly into all aspects of corporate networking. On the other hand, it was expected that the public networks would be very slow to adopt ATM because of the potential of leaving existing network investments under-utilized.

Over the past two to three years, the situation has changed dramatically. Private networks are using ATM in the backbone, but not broadly for desktop connectivity. Cost issues have precluded this from happening. Today and for the foreseeable future, the desktop connectivity will be based on Ethernet packets. On the other hand, carriers are introducing ATM very widely in their network because of its ability to deal with the bandwidth requirements of growing data communication and Internet applications. Indeed, it is expected that the public use of ATM networks will become even more widespread as soon as the use of switched virtual circuits becomes the norm.

This report starts by examining how we got where we are today, and then looks at what users are going to want in the future. It examines what ATM is, how it is being used, what its advantages and disadvantages are, and who is providing both services and equipment. It then projects the likely demand profiles and how that will impact revenues.

History of Network Evolution

Communications and the networks that support them have experienced major changes over the past several decades, but until now voice networks and data networks have largely experienced separate directions of evolution. Beginning in the 1950s, voice networks focused on providing universal service that allowed anyone to talk to anyone else with a single type of voice connection. This network evolved through a conversion to digital technologies and computer-controlled network management. The quality emphasis was on completing calls to their intended destination as rapidly as possible, while tolerating some slight degradation in transmission performance, which was readily accepted by callers.

A portion of the so-called voice traffic was actually communications between computers using modems running at speeds starting at 300 bits per second. In these instances, the voice network treats these connections as true voice calls and routes them in exactly the same way as voice traffic is handled. It is up to the modem to compensate for the transmission degradation so that the communication achieves the desired accuracy. Also, in these cases, the users' computers were responsible for determining if the data was complete and accurate.

As an alternative to dial-up modem connections, another data networking capability was built in the late 1960s called packet switching-based on a early military need for relatively well controlled communication between computer terminals. The protocol used, now called X.25, allowed computer terminals around the world to communicate in a manner that was completely different from that of the classic modem connection.

In parallel with the growth of the voice and packet networks over the past two decades, there has been an evolution of local data networking beginning with simple point-to-point connections of terminals to computers over dedicated private lines. Later, the introduction of local area networks (LANs) provided in-building connections between small computers. As the move to the personal computer (PC) accelerated, the need for LANs expanded dramatically, leading to the desire of users to intercommunicate between locations using the LAN protocols. As a consequence of this, a whole new network hierarchy was built to provide corporate data networking functions. These networks differed from the others in their reliance on very high bandwidth to compensate for rather simple protocols.

These three approaches (the voice network, the packet network and the local data network) have fundamental architectural differences. The voice network requires a connection to be established in advance of any communication, while a packet network relies on the address within the packet for routing the information to its destination as capacity allows. The local data network just sends information at will with destination addresses included in each message. This destination address is checked by all receiving terminals, with only the correct terminal capturing the message. The proper operation of a LAN relies on there being a lot of excess capacity to assure that most of the time, messages will be delivered properly and quickly.

The need to consolidate these networks became apparent as the demand for communication grew, even before the Internet became a factor. Thus, ATM was brought into the picture as a means of combining all of these communication methods together into one cohesive network structure.

ATM's Attractiveness

What is it about ATM that makes it so attractive? Why is it of interest across a broad field? What makes it the target of so many who want to maintain the status quo? The major reason is its flexibility in many areas, including:

• Medium—Overall, ATM has been architected to be flexible enough to handle a wide range of environments including almost all types of transmission facilities. Hence, ATM is fully compatible with most fiber optic transmission systems, most copper wire transmission systems, most coaxial transmission systems and even some wireless transmission systems.

• Distances—ATM has the ability to work effectively over very small and very large distances, 1 meter to 100,000 kilometers. Clearly, if it is to serve the needs of the long haul carriers, it must be able to go around the world. And yet, to serve the needs of the small business and the local network, it must work effectively over distances as short as just a few meters.

• Speed—One of the key elements that differentiates ATM from most other transmission protocols is its ability to work at a wide variety of speeds, from 10 Kbit/s to 1 Tbit/s. Almost all systems (LANs, long haul transmission systems, etc.) are specifically designed to operate at a particular speed.

• Media—Of course, the main attraction of ATM is its ability to seamlessly handle voice, data, video, and multimedia--essentially any type of information that's been identified so far--and to carry it through the network transparently. The result is a separation between the network that carries the information and the end elements that process the information, leading to much greater efficiency than if processing must take place as part of the network handling.

Network Development Alternatives

A few years ago, the expectation was that ATM would be the pervasive data and voice network solution partly because it was the only choice available to deal with the proliferation of network demands. Recently, several new approaches have been introduced to both the public and private networking arenas that appear to offer simpler ways to upgrade the capability of existing networks. Although the press likes to position these as competitive choices for solving the same problems, in fact, they each offer some advantages and some disadvantages in the expected environment of corporate and carrier networks.

The most likely network alternatives are:

• An Internet Protocol (IP)-based full packet switched network, converting all traffic (including voice) to Ethernet packets encapsulated in IP to be routed as needed.

• A frame relay-based network, working in a manner very similar to the IP network except the IP packets would be encapsulated to travel the frame relay structure, and pre-defined routes can be established.

• An ATM cell switching network, segmenting all frame relay and IP packets into cells and packaging groups of voice calls into cells, all using pre-established routes.

Insight's research into the ATM market indicates that IP, ATM, and frame relay networks, as well as combinations of the three, are currently being implemented in public and private networks. Nevertheless, ATM remains the strongest choice for public and private network interconnectivity, and it will be broadly used to merge the wide variety of information being transported. No matter what traffic format is used, they can all be carried on an ATM network. Of course, both native mode frame relay and IP networks can handle many of these formats, but not all. Fortunately, existing frame relay and IP networks can be retained because they can be easily combined under the ATM umbrella. There may never be a single solution network, be it IP, ATM or any other acronym-all will have to live together and be used where each is most optimum. All of these approaches have their places, and will all help move users to a new level of sophistication in services and connectivity.

Summary of Insight's Forecasts

Insight predicts that the Internet will drive data communications demand to far exceed voice. To accommodate this explosion in bandwidth demand, new networks have been developed and are being built. Equipment trends are driven by the growth in the ATM networks. Although there are some overlaps between private and public systems, the differences in size, service demands and reliability have led to systems being aimed primarily at just one of these markets.

The build-up of ATM will be a supply-push rather than a demand-pull market, with carriers using ATM for their own needs and creating an infrastructure that makes the users find ATM the most attractive way to accomplish their communications functions.

Back to Top