The Promise of ATM
Some of the old articles on communications technology that I've saved come in handy - not because there isn't more up-to-date literature, but because they remind me why, after almost thirty years in the field, I've become such a skeptic. I've seen so many promises broken. In the 1990's the communications gurus were certain that ATM was the technology that would bring about network convergence - voice and data, end-to-end, on one network infrastructure. Of course, it was not imagined at the time of ATM definition that voice would be delivered over IP. Just for "old times' sake", and maybe for a few laughs, let's recall a few quotes from the 1990's. I must point out that these experts' statements are consistent with the opinions of many of their colleagues at the time.
The following quotes are from Dan Minoli in Network Computing October 15, 1992. Dan's bio at the time stated that he was a full-time data communications strategic planner. He is the author of many books on networking, including some very recent ones. Referring to ATM as the "latest and greatest networking technology," he states:
"ATM technology clearly will be implemented in public networks, and corporate networks will connect to these public services in much the same way they now use frame relay or X.25 services." ... "ATM will also penetrate next-generation LANs at the core as a premises technology." ... "In other words, network protocols such as TCP/IP will need to be interfaced into the ATM data-link support. A user device could be a workstation configured with an ATM card, a router configured for ATM/cell relay service or another device such as a host or a video codec." ... "Future applications will require bandwidth on both the local and wide areas that other, non-ATM solutions just can't deliver." ... "Additionally, workstation vendors and carriers have realized there are mutually beneficial advantages to using the same technology locally and over a distance instead of using different technologies, as is the case today with LANs such as Ethernet and Token-Ring connecting with WAN links such as T1 lines. In fact, this is precisely the reason that relatively expensive netowrk bridges and routers are needed at all: They are to some extent fancy protocol and speed converters." ... "Note, however, that ATM will not necessarily eliminate the need for bridges or routers, but will diminish their primary functions in the protocol and speed conversion arena."
Mark Clarkson, who was a BYTE consulting editor when he wrote the following, was a bit more reserved in BYTE, August 1993. The article that he wrote quoted many experts in the field. Among his statements:
"Even in an industry renowned for hype and hoopla, it's difficult to comprehend the storm of hyperbole surrounding ATM. ... ATM is now being hailed as the key to interconnecting the world." ... "The truth is that the technology will do all these things, but not until all the pieces are in place." ... "Although ATM was conceived as a public telephone technology, LAN vendors and users are embracing it to such an extent that the majority of ATM products are expected to initially be used in private networks. Eventually, ATM will interconnect workstations and carry sophisticated multimedia applications, such as desktop video-conferencing." ... "Data can move from your desktop to a campus backbone, onto a private WAN and then to the public network, and back to a colleague's desktop. The same switching technology is used throughout the process." ... "ATM may be here, but as a standard it's still incomplete."
Jeff Gould was an independent networking consultant in 1994 (if it is the same Jeff Gould, he is now CEO and Director of Research at Peerstone Research) when he wrote the following conclusion in an article about ATM in Data Communications June, 1994:
"The one factor that ultimately may ensure ATM's long-term success is its ability to bridge the LAN and WAN worlds, a talent that sets ATM apart from its LAN-based rivals. Once data starts flying through the network at ultrahigh speeds, the fewer boxes and algorithms needed to sort that information out, the better."
So, what has happened and what hasn't happened?
All that was back in the 1990's, but it is now 2008, and the technology that is now being hailed - though definitely with less "hype and hoopla", perhaps because everyone now knows that no solution is the ultimate solution - as the infrastructure for network convergence is IP/MPLS. In April 2005, the ATM Forum, which was founded in 1991, merged with the Frame Relay Forum and the MPLS Forum to form the MFA (MPLS-Frame Relay-ATM) Forum. The new forum was organized "with the intent of converging the implementation of multi service packet technologies in the global networking communications industry," according to Andrew Malis, President and Chairman of the Board of the MFA Forum. However, it was decided that the name does not reflect their present focus and the forum was renamed the IP/MPLS Forum on October 22, 2007 to reflect their mission, as stated on their website, "to drive the global adoption of IP/MPLS-based technology, networks, services and solutions."
If the ATM Forum has basically evolved into the MPLS Forum, then one might wonder if ATM is a technology worthy of study in its own right. The answer is "yes!" True, ATM did not become the end-to-end infrastructure that was envisioned. Internetworking, that is, interconnection between different networking technologies, is still necessary, and probably will always be. And, therefore, routers are still required. ATM will apparently not be the technology to bring about convergence of voice and data. But, ATM has been widely implemented in public networks, and corporate networks do connect to these public services in much the same way they used to connect to frame relay or X.25 services. It will take some time before IP/MPLS replaces the ATM infrastructure entirely - after all, there has been a lot of investment, both monetary and in time, learning curve, etc., in the ATM infrastructure. Existing ATM hardware will serve as an infrastructure transport for IP/MPLS-driven networks. And until businesses are convinced that Voice over IP can provide business-quality voice service, not all businesses will choose to migrate voice services to an IP infrastructure.
Perhaps most importantly, understanding ATM is a foundation for understanding IP/MPLS. IP/MPLS was not conceived in a vacuum, but is derived from experience with and lessons learned from ATM. The concepts of label-switching are basically packet-switching concepts, and ATM is a packet-switching technology. Perhaps, if IP/MPLS lives up to its expectations, then in several decades IP/MPLS will be explained devoid of ATM terminology. But you'll probably have to wait until the old-timers like me (and even younger folks) disappear from the scene. In the meantime, we still talk of X.25, though most of you who read this will not ever have experienced it, just as we talk of the latest last-mile solutions in terms of dial-up links, which many of you may never have experienced. (If you ever used an acoustic coupler, raise your right hand. How about a slide rule?)
ATM Tutorials
On the menu to the left, you can find a tutorial on ATM and a super ATM interactive applet - recommended even if the technology is useless. There is also a packet capture emulation program to help you understand ATM message formats. The ATM tutorial includes a page on ATMARP, a protocol that was defined as part of the Classical IP over ATM standard. This portal also has a link to the IP/MPLS portal. Once you understand ATM, you certainly should not stop there but check out the next stage in network convergence.
Thank you for visiting RAD University. We hope that the tutorials will be helpful and we welcome your comments.
Debby Koren, "Dean" RAD University
