Standards and Consortia, More Valuable than Ever
It’s an exciting time in the embedded industry. Lots of innovation is happening in serial communications, such as PCI Express and RapidIO; in wireless networking, such as WiFi 802.11g and WiMax; and in hardware choices, such as mini-ITX, EPIC, ETX, and even VITA-46 (I know, this is the wrong magazine to be talking about VME but it serves to make a point). Take a look at the consumer side and you’ll see lots of choices but no platform convergence in games consoles, cell phones, or MP3 players (can anyone topple the iPod?). From the X-box to the PVR to the PDA, and right down to the circuit boards inside these devices there’s a lack of consensus over physical size, I/O connectors, bus type…just about everything is different. Only the ICs show some commonality.
Diversity is good, to a point. Manufacturers crave it because it makes their products unique. But “sameness” – at least inside the box – drives down prices by increasing commodity status. And interoperability benefits users as information and data is shared between widgets, expanding functionality beyond a single device. The innovation explosion is proliferating “difference” at a rate that can’t be good for users in the long run. Platforms become standalone islands, and the entire user experience, familiarity, and system functionality is discarded when the hardware breaks or becomes obsolete. It’s a huge problem on the consumer side, and our general embedded market is showing similar signs of proliferating “difference”.
I’ve been in the embedded industry for a long while, and have gone on the record many times in favor of industry open standards, trade associations, and consortia. The PC/104 Consortium, the standards body that acts as the keeper of the sacred PC/104 flame, understands fully the importance of open standards and innovation. That’s why the group added EBX and EPIC to the list of official PC/104 Consortium standards over the last 18 months. But the Consortium is under increasing pressure to innovate along with the embedded market, while fighting off the proliferation of proprietary small form factor hardware sizes. (We’ll take a comprehensive look at these in the Fall issue.)
I was recently contacted by the RapidIO Trade Association (www.rapidio.org),conducting a survey on the value of open standards. My response to them went like this:
Open standards often create markets, whereas proprietary implementations (sometimes called "standards" by the companies who develop them) stifle the creativity of the larger market. When a company creates some IP and opens it up to the world and the company's competition, additional "brain power" can be applied to making that standard more useful while applying it to many more markets. Often, the standard finds its way into applications and markets that the originator never intended or dreamed.
A company that maintains a proprietary implementation (a closed "standard") can garner 100 percent of their total available market (TAM), depending on how they define their market. But once the standard becomes open, and more vendors jump on board, a company (including the originator) will definitely garner less than 100 percent of the TAM, but the TAM is usually many, many times bigger. The situation goes from 100 percent of a small pie, to a smaller slice of a much larger pie. This is not just because new opportunities are found, but also because customers are more encouraged to use a standard once there is competition, numerous vendors to choose from, and a growing network of support products and services. Few customers want to rely on proprietary and closed sole-source IP "standards".
For example, AMD's proprietary TAXI chipset was intended for high speed serial communication to replace bus architectures. The revolutionary at the time (late 1980's) TAXI (Transparent Asynchronous Transceiver Interface) used 4B/5B encoding with 8-bit interfaces at either end, but a 2-wire link (simplex) in between. Despite all the excitement about 100 Mbps serial communication (at 125 Mbaud), the chipset never achieved wide market success. It wasn't until AMD offered the technology to the Fibre Channel ANSI committee - and opened up the "secret sauce" a bit more - that the technology found a real home.
Did AMD invent modern serial communications technology? No. But the TAXI - once its proprietary moniker was removed - showed the market and myriad other companies that the concept of serial communications was valid and could be realized. AMD eventually actively promoted the TAXI in the original ANSI committees for FDDI, Fibre Channel, HiPPI, ESCON, and others. The TAXI (now using 8B/10B and running at 166 Mbps) was often used to prototype the fore-runners of myriad serial standards such as RapidIO, InfiniBand, PCI Express, and many others.
In short: AMD's small market TAM for "bus extenders" has burgeoned into the hugely growing market for serial communications - most of which are based upon the early proprietary principles and implementation of the TAXI chip set. Open standards create markets: they're the right approach.
I have personal experience with this situation, having managed the TAXI and Optics group at AMD and initiated the strategy for expanding the TAXI into open standards. Innovation is critical in our industry, but we need to remain conscious of the need to create interoperability through standards. Industry trade associations and consortia are the right places to balance innovation with open standards.