New consumer technology does not equal ideal COTSStory
June 24, 2008
Today's consumer technology is so advanced and alluring that we're all primed psychologically to expect it in all applications - even complex COTS-based military systems. This priming subconsciously makes us gloss over the fact that consumer technology was not designed for high-performance defense applications and could result in serious repercussions for the COTS industry.
At one point, much of high-technology development and deployment started in the military and aerospace domain and only later filtered down to commercial products. With the decision to go COTS and the obsession with a device-driven consumer market in the past 5-10 years, this trend has reversed. More often than not, the mil/aero market is playing catch-up with consumer-craze technologies as customers demand such technology in their systems. Overall, the adoption of consumer commodity technologies for mil/aero applications would seem to be the ideal realization of the COTS initiative. And yet if we're not careful, I believe it could jeopardize customers' trust in COTS products.
Our gadget- (PDA, MP3 player, thumb drive, handheld GPS, and so on) and connectivity-obsessed world is rapidly pushing the evolution of technology. The latest and greatest technologies often turn up in consumer electronics before being migrated into the complex embedded space. And so, generations of x86 CPUs appeared in home PCs and servers well before they made their way into embedded form factor SBCs. The performance level of some elements of consumer technology is so high that home gaming systems like the Sony PlayStation 3 have to be controlled under the U.S. Export Administration Regulations in order to prevent its leading-edge technology from falling into the United States' enemies' hands.
Surprisingly, while consumer technology's lead over embedded should herald the ideal realization of COTS, it might have a negative effect on the COTS world and, consequently, end customers. The reason: The consumer technology craze has greatly affected people's perceptions and expectation of what technology can do. To borrow a concept raised in Malcom Gladwell's nonfiction book, "Blink," we all have been subconsciously "primed" by the level of daily technological exposure.
When you can hold 2 to 4 GB of storage in a 2" x .5" USB flash drive or an iPod shuffle, you're psychologically primed to expect that very technology in your defense or industrial applications. These expectations affect us at a visceral level, before our analytical faculties have considered whether the technology is truly a suitable technical match. Therefore, the "priming" presents openings for new entrants in the COTS marketplace to begin nailing chips to boards, without fully considering the application differences. The ill-suitability of these products would only become evident when customers are farther along the course of system integration. The ensuing frustration could jeopardize the collective equity of trust in COTS products that vendors have been painstakingly earning from their defense customers since the push towards COTS began in the 1994 "Perry Memo."
Let's take this a step further and see how this trend affects high-performance interconnects. Ethernet is pretty much the dominant networking interconnect above the board-to-board level in the commercial and defense space. Even at the board-to-board level, most recent COTS products already have the facility to use 100 Mb or 1 Gb Ethernet for control-plane functions. Ethernet's ubiquity and enormous supplier base make it an obvious example of ideal COTS technology for the control plane.
Now, at 10 Gbps speeds, Ethernet is fast enough to be used for the defense application data plane, too. Its combination of speed and bidirectionality make it very attractive as a system interconnect and for linking high-speed sensors to processing, as an alternative to more specialized interconnects like Serial FPDP or Fibre Channel. So, it appears we are on the cusp of a period when commercial interconnect and I/O products can meet high-performance defense application interconnect requirements.
Given this, one would think that 10 GbE technology would be an example of ideal COTS technology for the military. However, pause before rushing out to grab your crowbar, prying silicon off of the nearest 10 GbE network interface card, and bending it into a shape that will fit your defense applications. You may be "primed."
That Ethernet silicon is designed for mass market data-center applications, where servers and PCs containing large amounts of memory are communicating in a non-real-time or soft real-time setting. Contrast the mass market applications to high-performance defense applications that are subject to hard real-time constraints, sensors with limited (if any) memory, and no opportunity to resend dropped data, never mind requirements like microsecond-accuracy data time-stamping and optimized bridging to other protocols. While the 10 GbE standard is well suited to meet these requirements, the mass market silicon will vary from being merely challenged to being stumped trying to meet them.
In conclusion, although it's tempting to think that we are reaching a point of ideal COTS, when consumer technologies can ubiquitously be applied to the mil/aero sector, the unique needs of defense applications should ensure a place for COTS vendors going forward. Emerging mass commercial market interconnect standards and technologies can meet the requirements of high-performance defense applications, but the mass market implementations themselves, having been designed for a different set of requirements, typically won't be suitable. Designers, developers, customers -- in fact, all of us -- need to be conscious of the effect of psychological priming. We should also think carefully before assuming that commodity products will become the ideal COTS products.
Rob Kraft is VP of marketing at AdvancedIO Systems. He has more than 12 years of experience in systems engineering and business roles in the embedded real-time space. Prior to joining AdvancedIO, he worked at Spectrum Signal Processing and AlliedSignal Aerospace. Rob has an MASc in Electrical Engineering from the University of Toronto. He can be contacted at [email protected].
AdvancedIO Systems Inc.