When launched a decade ago, the Advanced Telecommunication Computing Architecture (ATCA) was heralded as the way to make next generation communications equipment available to the market without the risk of using or developing proprietary technology.
ATCA was the result of the largest development effort made by PICMG, the PCI Industrial Computer Manufacturers Group, with more than 100 companies involved in the process. While, as the technology's name suggests, it was developed for communications applications, it has found supporters from the industrial world. But while the technology hasn't broken through in quite the way its supporters might have thought, efforts continue to adapt ATCA to the industrial world.
Rob Pettigrew is marketing director with Emerson Network Power's Embedded Computing business. "There has been strong adoption of the technology in the comms sector. Most carriers have adopted ATCA, with one notable exception, and they see the advantage of the technology being available commercially, rather than being proprietary. They also see advantages when it comes to cost and life cycle."
Pettigrew noted that ATCA has seen three main iterations in its history; from its original support for 1Gbit Ethernet, ATCA has moved to support 10G and, more recently, 40G comms. "These technology insertions have opened new markets and applications," he contended. "The 1G and 10G variants were applied in the control plane, but we're now seeing a big investment in 40G, which allows ATCA equipment to be deployed in the data plane to handle tasks such as packet processing."
ATCA was developed to enable companies to move forward without making significant infrastructure investments. "If you think back to the 1970s," Pettigrew suggested, "telecoms companies only delivered a dial tone and there were no time to market pressures. Now, they are under tremendous pressure to provide more and more bandwidth and to keep up with demand. ATCA is helping them to do this."
Supporting the full sized ATCA processing boards or blades are Advanced Mezzanine Cards, or AMC modules. Although intended for use in ATCA applications, AMCs have found applications elsewhere including enabling the MicroTCA format. The reason? AMCs are powerful enough to standalone, needing only to be plugged into a suitable backplane.
MicroTCA, which first saw the light of day in 2006, has had a mixed reception. Early in its life, the approach was seen as a potential solution to problems which VME, for example, couldn't address. Today, some think MicroTCA has been sidelined to some extent.
Irene Hahner is MicroTCA product manager with embedded computing specialist Kontron. She said MicroTCA is typically used in industrial systems where there is the need for high computing performance and high bandwidth. "The technology has been well received by the applications to which it is suited; one example is image processing systems. The MicroTCA comms path uses Serial RapidIO (SRIO) and image processing applications typically feature dsps and fpgas which are attached using SRIO."
Peter Ahne, a Kontron product marketing manager, added: "Cameras are only one input; there are systems in which other forms of sensor deliver data that needs to be processed rapidly. These types of system not only need high communications bandwidth, they also need a scalable processing capability."
Ahne also offered metal foil production as an application well suited to MicroTCA. "You need a control system which measures the thickness of the foil being produced and then adjusts the pressure being applied by the rolls. Because foil production is a high speed process, the thickness needs to be measured and the roll gap adjusted very quickly. This fast control loop is suited to MicroTCA."
But Pettigrew isn't particularly taken with MicroTCA. "While ATCA has a strong future, that for MicroTCA is not quite as rosy." His reasoning behind the statement? "I think MicroTCA is a standard that was invented without a clear market focus. AMCs were conceived as mezzanine cards for ATCA applications, but someone said why not plug them into a backplane? Having said that, Motorola Embedded Computing (acquired by Emerson) did make a big investment."
Pettigrew sees MicroTCA failing on two counts. "Low end applications don't need serviceability; if the equipment fails, you replace it. At the high end, you can't build high performance, high availability MicroTCA systems because you are constrained by the modules. I think the problem is better solved by CompactPCI."
Hahner conceded there is competition for MicroTCA. "It does compete against other form factors in the industrial market; for example, we do see a lot of box pcs being applied. But MicroTCA does offer a number of advantages, including remote management and the ability to build redundant systems easily. And it does support SRIO, along with 10G Ethernet." She also pointed to the experimental physics sector, where MicroTCA has been selected to replace VME systems.
Pettigrew does, however, see a use for MicroTCA: prototyping. "If you need to build a system quickly a basestation, for example you need an fpga array, dsps and packet processing. You can buy these boards, put them into a system, get an architecture developed quickly and start developing software. But you wouldn't deploy that system."
Bearing in mind Pettigrew's comments, it will come as no surprise that Emerson is no longer investing in MicroTCA. And neither is PICMG currently considering any developments. In the most recent improvements to the MicroTCA specification, a hardened conduction cooled module (MicroTCA.3) was defined to meet more stringent environmental conditions, while MicroTCA.4 defined an additional rear I/O area; something pushed by experimental physicists.
Ahne pointed out that MicroTCA already has two levels of ruggedness and that some industrial automation companies were looking to take advantage of the features of MicroTCA.4. "At the moment, we don't see the need to extend the spec," he said.
ATCA, meanwhile, looks to be recovering some ground lost over the last couple of years to other approaches. Three working groups are looking to develop the specification; two of these are addressing issues raised by the experimental physics community, while the third is looking to extend the applicability of ATCA beyond the telecom central office.
However, both ATCA and MicroTCA have one target market in common military. Hahner said: "We see significant market growth for MicroTCA in a range of segments and military is one.
Many military applications have used VME in the past and many of these are moving to the VPX platform. That's a natural development, but MicroTCA can compete in some of these cases, bringing advantages such as cost."
Pettigrew, meanwhile, sees ATCA being adopted in military systems, if only slowly. "Areas such as radar are benign computing environments which are similar to telecoms, so ATCA works well in them. We're targeting those comms intensive applications which use Ethernet, but the barrier with military apps is getting ATCA adopted in the first place."
Yet MicroTCA may not have reached a dead end. "Technology continues to develop and we will need to review the situation in a couple of years," Hahner concluded. "But for the moment, MicroTCA's features are sufficient."
