ATCA: Ready to Take the Stage

Fractal Realms series. Backdrop of fractal elements, grids and symbols on the subject of education, science and technology

The Advanced Telecom Computing Architecture (ATCA) is moving into the real
world. Beyond trial installations now, Huawei, the Chinese telecom company,
is deploying real systems to serve real customers in Q4 of this year. Korea
Telecom recently completed a pilot project deploying ATCA that was accomplished
in half the time with half the personnel as previous efforts. The recent Supercom
trade show in Chicago had some 50 booths demonstrating ATCA-based technologies.
The questions today are not whether ATCA will catch on, but rather, where, how
fast, how big and in what areas other than telecom?

ATCA should really be thought of as more than a great-big form-factor backplane
board standard. It is the result of concerted efforts by standards organizations,
as well as hardware and software companies in a variety of arenas, some of whom
are definite competitors. The ATCA, or PICMG 3.x specification, comes out of efforts
of the PCI Industrial Computer Manufacturers Group. A large group of companies
including Intel, Motorola, Radisys, Artesyn, SBS, Force and ZYNYX, to name a few,
are designing and building ATCA boards. Over sixty companies in all are producing
boards, components, chassis, backplanes, connectors and other components to support
the standard.

On the software side, the specifications created by the Service Availability Forum
(SAF) are defining application, middleware, hardware and management interfaces
that are directly applicable to ATCA. Under the aegis of Open Systems Development
Labs (OSDL), carrier-grade Linux is being tailored to the ATCA environment. Other
companies such as GoAhead Software and Eternal Systems are producing high-availability
products. These are only representative and their mention is meant to exclude
no one.

One thing that ATCA has done is raise the concept of what we think of as a “platform.”
There was a time when platform referred to a combination of processor and RTOS.
Well, it was a piece of hardware with an API after all. A platform is the point
at which you have to start in order to produce a product that is of value to a
customer. Often times that has meant doing a lot more development work and integration
than will actually be noticed as unique by the end user. This includes things
like board support packages, peripheral integration, driver development, packaging
and the like.

By contrast, when you put together all the components that have been developed
to support ATCA, you come up with a chassis that includes power supplies, fans,
pre-thought-out thermal management, a backplane and connectors, shelf management
hardware and software, a mix of boards for the application domain, a set of standard
software interfaces, an operating system tailored for the environment and soon
the ability to design or buy more specialized hardware and software in the form
of AMC modules.

This level of prepackaged integration may have a bit of a double edge to it, however.
Integrating boards, physical infrastructure, system software, management hardware
and software still requires effort and verification. However, at a certain volume,
once that work is done, the base system at the purpose-built box level almost
becomes a commodity. At that point economics of manufacturing kick in and the
ability to play at this level might very well be limited to large manufacturers
with large customers.

The danger for even these large suppliers could be that large customers take the
next logical step and bypass the OEMs, doing their own basic integration and going
to contract manufacturing. At any rate, the business models of a number of traditional
equipment manufacturers stand to be profoundly affected by such developments and
it will be interesting to see how that all plays out.

Does this mean that the world of ATCA will become the exclusive playground for
a very small number of very large companies? Possibly, if the activity is limited
to supplying pre-integrated chassis and ATCA boards. However, there is a possibility
that niche markets may open up in either specialized areas of communications or
in non-telecom-related application areas. This will provide an opportunity for
smaller, innovative companies to participate with targeted designs on AMC modules.

In this sense, think metaphorically for a moment of an ATCA carrier card representing
a previous generation backplane chassis that has been smooshed down to the form-factor
of a single large board. Depending on power consumption and size, you can get
up to eight AMC boards on one ATCA card. There are many contexts where that can
constitute an entire system—and a sophisticated one at that—on a single
board.

If economies of price and performance seem attractive, there could be a wealth
of opportunities awaiting those who can identify the right combination of ATCA-related
components and technologies to solve problems in instrumentation, industrial control,
security, aerospace, transportation and many other fields. We are still too close
to the beginning to be able to see clearly but the future looks hopeful.

The Williams Law of Technology Application clearly states that the original
innovators of a basic technology have utterly no idea how that technology will
ultimately be used. That may turn out to be the case with the universe of ATCA.
“But we thought we were going to revive and transform the telecom industry,”
they will say. The rest of the world will probably say, “Surprise!”