Embedded systems obviously is drawing a crowd, causing us to spend a great deal of time and effort worrying whether JavaOS, Win32 Real Time or Inferno are serious threats to WIND's leadership. All the recent posts about Inferno, for example, show proper thinking about the problem, taking the treat seriously but not panicking or making hasty, silly conclusions about exaggerated impacts.
Inferno is a classic case of "too little, too late". Actually, Bell Labs tried to push Plan 9 over a year ago as a network-oriented operating system, and that never got off the ground. (Almost one year ago exactly someone on an Oracle thread was trying to convince me that Plan 9 would be used by Oracle for the NC OS.) Inferno seems to be a rapid re-do of Plan 9 to incorporate Java-like features. It is yet another virtual machine that promotes yet another C-like language. That the language is not object-oriented is a major reason why it is too little. Trying to follow on the coat-tails of Sun's massive effort to sell Java is why it is too late by at least two years.
Although Lucent is no doubt more agile than AT&T before the split, there is nothing about the company that suggests it has any capability whatsoever in the difficult world of computer hardware or software. While it is true that Bell Labs is the father of C, C++ and Unix, the lab did little commercially with any of them. AT&T never even knew it owned Unix until well into the 1980's when it quickly formed USL to shepherd its development and issued statements about how Unix was a trademark. It partnered with Sun to lead the parade, only to discover that nobody was following (IBM, HP, DEC, et al) didn't want to play in the AT&T/Sun ballpark and split off forming OSF, successfully challenging Sun (which separated quickly from AT&T) over the look-and-feel of Unix. Exasperated, AT&T anointed Novell as keeper-of-the-Unix-key, which lasted through most of one of Novell's visions. (Incredibly, Novell sold the key to SCO for safekeeping.)
If JavaOS is not a threat to VxWorks, how in the world can Inferno be? Not one networking company could possibly risk depending on a major competitor for the heart of its networking products. With all these companies proscribing the use, not only of Inferno, but Limbo, the new programming language, what chance is there that Limbo can become a standard? Less than none. Languages that fail to achieve critical mass through popular use nowadays die immediately.
Since this is all so clear to all of us on this thread, Lucent must see it also, so you are probably wondering why Lucent is wasting resources developing an unneeded Inferno and Limbo. What else can they do with that group of gray-hair, emeritus programmers still hanging around from the ivy tower heydays of Bell Labs? They shoot horses, don't they?
Lucent may choose to use Inferno internally in their many networking products, but if they do it would be a critical mistake. Lucent should just drop Inferno/Limbo and go on. The days of roll-your-own RTOS are over, no matter how good they are, and successful companies, including networking companies, are the ones that stay in the mainstream of open systems.
Back to JavaOS as a potential threat. There is no doubt that Java will be big, but it will never push well-established languages like C, C++ and Assembly Language out of the embedded systems market place. For example, the large majority of internet applicances can and will be built without using Java. Java will be the language of choice only when multi-platform, dynamic program updates are needed often, most notably where flexible user-interaction is required, as in the PC paradigm. The lion's share, by orders of magnitude, of future internet appliances will not need highly dynamic updates, obviating completely the requirement for Java. Once Java is ready for prime time, however, with so much use where it is needed, it will slowly gain market share as a generic programming language even where it is not needed per se.
This means that, on a percentage basis, very few internet appliances, including the NC, will be willing to live exclusively with Java. This is a major reason why Sun's attempt to capture the embedded systems space with a JavaOS processor is flawed. Real time is another. And for both of these reasons, the best way to provide JavaOS is as a layer on top of a robust RTOS. In this regard, INTS and MWAR remain WIND's most serious competitors, not Sun.
But this leaves the elephant, Microsoft. What about the key-note speech at the Embedded Systems Conference in which a General Manager was to present startling information about Win32 Real Time. Well, word is it was a non-event, I suspect even disappointing David Stuart. The elephant still sleeps.
I couldn't get up to the conference to enjoy the speech, but I did slip in today to check on how it was received. I talked to the people in the Microsoft booth. The speech turned out to be a pitch for programming in Win32 API, noting that PC's are used in numerous embedded applications, and there are various third party real-time extensions to Windows NT. Microsoft was pushing Windows CE, and continues to expect many devices will come to market using that OS. The unit cost of Windows CE probably lies in between that of MSDOS and Windows 95 through the regular Microsoft distributors, but for numbers exceeding 20,000 you can expect to negotiate directly with Microsoft. I tested the Microsoft representative by referencing the Bill Gate's quote in a Time magazine article that appeared about a month ago - and he recalled the $10 unit figure without a prompt. He also acknowledged that not only was Microsoft-At-Work DOA, but the people on the team had been incorporated into the Windows CE project. Lastly, he confirmed that Microsoft was rumored to be developing an RTOS, but he knew nothing specific about it.
While Microsoft has the theoretical potential to be a serious competitor in the RTOS space, they still have not made significant overt moves that are particularly threatening. Their trump card, the Win32 API, has been finessed by the Willows RT announcement. To the extent that the Win32 API is popular or that Win32 API legacy code exists, it can be accommodated within VxWorks via Willows.
Meanwhile, the umbrella of capabilities and third-party development tools comprising Tornado continues to strengthen rapidly. INTS demonstrated pRISM+, but strangely without showing anything about pRISM+, just pre-existing components like ESP and Single-Step. INTS is seriously behind the eight-ball in regards to competing against Tornado, and perhaps they would be better off just canning the pRISM+ idea. Microware announced an inexpensive IDE developed with Metrowerks, but surely something as important as an IDE should not be shared with an outside firm. Unfortunately, I didn't get a chance to check it out at the show.
As we look at how things have changed over the last year, it is clear that WIND's position has strengthened significantly compared to its peer group. Microsoft is still a non-serious RTOS player, and no other serious contenders from the land of the giants have surfaced, albeit not without attempts by Sun and Lucent. I2O is meeting every expectation, and should single-handedly propel WIND to much greater heights. While design wins in the telecommunications, networking space are impressive, we can expect wireless design wins to explode starting this year. Unlike many other application areas, wireless is so new that there are no entrenched OS's, roll-your-own or other commercial, that need to be displaced to make the win. Consequently, WIND is certain to garner at least their share of a an explosive aspect of the embedded systems market.
Clearly the market is not as sanguine about fending off all these threats as am I. WIND's stock price continues to leak off without any apparent reason, other than all the many things talked about on this thread over the last couple of months: Barrons, Win32 Real Time, Inferno and JavaOS. In a lapse last week, I labeled the market crazy, and was quickly rebuffed for improper deference to the Market God. (And was surprised to learn not only of its existence but also about the one true Bridge God, which seems to behave exactly like the Golf Gods with which I am intimately familiar - and fear with good reason.)
Since the market never doubted for one moment that WIND would report a stellar quarter (which it knew would happen given the timing of the split announcement), it is either bothered about longer-term potential for disappointments or it is just not willing to pay appropriately for any potential from a small stock when safe money can be made in blue chips. Since there are many other high-tech stocks being treated similarly, I strongly believe the latter explanation applies. (Actually, I think there has been a major correction in small, particularly high-tech, stocks which is clouded by the continued advance of the major indexes.) In either case, it is an error, or market inefficiency, which can be profitably arbitraged.
Daily price volatility and volume suggest that the current tempest bothering the stock is on-going, so even though we may believe the market is shortsighted at the moment, it is not clear what lies ahead in the near term. Fortunately, the vagaries of the long-term only pertain to the extent of dizzying heights yet to be achieved.
Allen |
