>Do you have a comment on Wind River's entry into the Embedded Internet Device
>(EID) market? When do you think any of the EID's will actually be sold?
I would like to provide an extensive answer to this question - because of the over-looked importance of WIND's announcement.
Recall from early posts our view of sophisticated (32-bit MCU) embedded systems as the 3rd major wave of computer processing. Mark Weiser calls this ubiquitous (hidden) computing. Ubiquitous computing began to occur noticeably in the beginning of the nineties, and probably by 1996 surpassed the PC paradigm (the 2nd wave of computing) in units shipped. By the turn of the century, ubiquitous computing will be at least 10 times greater in numbers than desktop computing. Indeed, ubiquitous computing will invade the desktop (a la I2O, distributed computing components e.g. disk drives, keyboards, TV, etc. all deploying hidden computers).
While uses for intelligent devices, such as washing machines that monitor the water, or clothes dryers that monitor when the socks are dry, are numerous, and easily will out strip traditional use of computers in the mainframe or PC paradigms, this is, relatively speaking, very small potatoes. Even if all these devices require 32-bit processors, and exotic fuzzy logic with various sensors, and even though they will bring $millions in royalties to WIND, smart devices like these will not prove to be a significant part of future embedded systems. Neither will the NC, or even I2O. Since I2O alone justifies the current price of WIND, these statements may come as a surprise to anyone following the intense discussions surrounding that wonderful opportunity. But the fact is, when you project the potential numbers of these products against the really big numbers of important uses of hidden computing that are now able to emerge, all these applications of embedded systems pale in comparison.
What is the really big set of applications that will dramatically out-number I2O, the NC and so much more? WIND's announcement Monday contains all the clues you need to begin to appreciate where embedded systems is really going big time, and when. It is ironic that such an important announcement goes without fanfare, only causing a small blip in the stock price. Perhaps the Embedded Internet Device (EID) just doesn't seem any different than anything else embedded, or other uses of the internet? Nothing could be further from the truth. (The NC is arguably an EID, but it lacks relative signficance because it is only one such device; generic EID will contain hundreds of thousands of types of devices.)
The announcement said that WIND has packaged all the necessary software components needed for software engineers to rapidly develop customized client/server applications, where the client is any type of intelligent device with communications capability, and the server is simply any old computer connected to the internet, programmed to respond to the client. Coupling ubiquitous communications with ubiquitous computing is what makes EIDs such a rich concept.
Connectivity between the client and the server utilizes the available, zero-cost internet as a world-hopping communications leg. The client can connect to the internet via wire or wireless, again using communication components made available by WIND.
This tedious-sounding announcement means that any run-of-the-mill software engineer with a little hardware engineering support can utilize combinations of these software components (flash memory file system, graphics, window interface systems, including internet browsers, communications using any popular protocol and mechanism, and no doubt much, much more) to assemble a custom application very quickly. WIND will even provide guidance concerning processor selection in already packaged boards that fit needs of various types of devices. This unleashes hundreds of thousands of application systems programmers who can now undertake customized, client/server applications for commercial production or private company use. These applications can be completely free of the traditional PC, relying instead on ubiquitous, hidden computers. Unlike the NC, STBs, Digital TV and even I2O, the infrastructure is 100% in place today to support systems involving EIDs, where connection to the internet occurs via PCS, cellular, copper wire telephone or LAN. Soon LEO satellite links will be available to broaden mobile connectivity beyond metropolitan areas.
Although Java support is offered, my guess is that most EIDs today will be constructed using the C/C++ language, because most of them will have little need to constantly alter programming logic, which is what Java is particularly good at enabling. The point here is that there is no need to wait until Java matures to develop most EIDs.
The number of such applications can be expected to rocket to Mars, passing Pathfinder on the way. Naturally this will lead to a surge of Tornado development licenses. The number of clients associated with EACH application will vary from tens to thousands to millions, leading to uncountable (but no excuse for not trying) numbers of royalty-paying targets.
MWAR also is offering a packaged solution to this problem, but the biggest prize goes to the market leader.
Allen |
