May 2003

ATCA Revisited

BY JIM MACHI

A few months ago in this column we discussed Advanced Telecommunications Computing Architecture (AdvancedTCA, or ATCA). I subsequently received a few reader e-mails and also engaged with industry colleagues in some intense discussions on this subject. Having always planned to do a follow-up article, I�ve decided now is the time after a recent industry event where I met a reader who wanted to talk to me about �that ABCTA thing.�

To recap, ATCA, developed by the PCI Industrial Computer Manufacturer�s Group (PICMG), is targeted as a way to define a standards-based, high-availability platform for hosting carrier-grade telecommunications applications. This means good targets for ATCA product lines will be those that require things like faster and more capable silicon, increased I/O throughput, better cooling, increased availability, and faster internal routing. You can learn more about ATCA and the specs in my September 2002 column or by visiting either the PICMG Web site or the Intel Web site.

An issue readers had with my original ATCA article was how I could write about something that didn�t yet exist. To me, that is not an issue. Yo -- the name of this column is �Industry Insight!� One of its charters is to prepare you for the future. Writing about an important industry initiative like ATCA, which is relevant to Internet telephony, certainly fits into that mission. In the past I have written about SIP and MGCP in the same way.

Besides, the ATCA specification is now ratified (which it wasn�t last time we discussed it -- hence the objection). There have been, and continue to be, interoperability workshops among the PICMG members developing ATCA products. There have been three ATCA plug fests, with another planned for May 19 through 23 in Las Vegas. Press will be invited to that one, so I expect you�ll soon be hearing more public pronouncements from firms developing ATCA products.

What I always wanted to follow up on, and what the fellow at the tradeshow wanted to talk to me about, was that he �just didn�t get it� about ATCA, as he put it. He didn�t understand the need for a standard to supplant CompactPCI.

This is a common misconception because, in my opinion, the ATCA standard doesn�t supplant CompactPCI. It�s there to address a different market segment. Many readers of this magazine are comfortable with -- and support -- a standards-based computer telephony integration (CTI) world in which Internet telephony now plays a big part. But in the world of DSL/cable/wireless/PSTN central office communications, most equipment is still proprietary. ATCA addresses that market segment by using standards-based building blocks instead of the existing vertically-integrated equipment. And, based on experience in the computing and CTI worlds, this approach in a market segment proves over time to drive several major benefits.

The first key benefit of using standards-based building blocks is that it drives lower-cost price structures as ecosystems develop based on price/performance. Also, when vertically-integrated systems are built with standards-based building blocks, they do not need to be rearchitected every time a change is required. Only part of the architecture needs to be reintegrated, which improves time to market and provides greater flexibility to deliver service offerings. This model has proven successful in other market segments; it will ultimately prove successful in these new market segments.

Another benefit of ATCA for Internet Telephony� readers is that it inherently supports packet-based infrastructure from the ground up. That is, Internet telephony is inherently easier -- and will actually be the norm -- for these systems, since they are packet-based to begin with, with special techniques required to deal with TDM (time division multiplexing), a common method of traditional voice traffic.

One example of this I wrote about in the first ACTA column was ITDM. That�s IP telephony progress: a developer now has to go out of his way not to deal with packet telephony. Why? Because the backplane options are all packet-based. As such, packet-based infrastructure equipment (like equipment handling ATM traffic and Ethernet) makes ideal solution set targets. This would point to such infrastructure as 3G RNC (Radio Network Controller), 2.x G BSC (Base Station Controller), and xGSN (GPRS Support Node) in the wireless segment; multi-service switches in the routing segments; media gateways and softswitches in the voice traffic segments; CMTS (Cable Modem Termination Server) in the cable segment; and storage controllers in -- you guessed it -- the storage segment.

I�m sure this article addressed some lingering ATCA questions. But I�m equally sure it created new ones. If so, I�ve done my job. Because that�s what this column is all about.

Jim Machi is director, Product Management for the Network Processing Division of the Intel Communications Group. Intel, the world�s largest chipmaker, is also a leading manufacturer of computer, networking, and communications products. For more information, visit www.intel.com.

[ Return To The May 2003 Table Of Contents ]