This illustrates how WIND can compete aggressively for set-top boxes against MSFT or anyone else. It is also an update about Oracle's travails with the NC. But first, some history:
We tuned into the NC excitement in the fall of 1996 when WIND announced Oracle had selected VxWorks for a 2nd generation NC (presumably to be delivered about now). In checking out NCI's published plans then, I discovered Oracle planned to implement the NC consistent with their imposing "NC Architecture". That cemented by instincts that the biggest problem, and the biggest opportunity for Oracle, was on the server side. It also caused me to suggest that Oracle needed to develop a consumer version of the NC, where legacy applications (a major complicating feature of NC Architecture) would be non-existent, giving them a chance to get something in place during our lifetime. Later, I was even more emphatic about expediting a consumer version in order to keep the pressure on MSFT. I realized that LAN bandwidth was sufficient to enable MSFT to defeat every important benefit of the corporate NC with PC's, well within the time it would take to implement the NC Architecture. On the other hand, consumer bandwidth limitations effectively precludes MSFT from defeating most NC benefits for consumers, at least until broadband connectivity is generally available. This would give Oracle the window they need to grow the NC market. Meanwhile, this fall everything seemed to be at a standstill with respect to the 2nd generation NC, changing our view of NC royalties expected to accrue to WIND.
Now, let's get up to date. I think the infamous Network Computer Architecture is history, because I can't find any sign of in NCI's homepage. NCI has formally categorized the NC into Corporate and Consumer versions - as suggested. However, their definition of a Consumer NC has changed, and changed a lot. In fact, it has changed the whole picture. I think WIND comes out way ahead with the change (certainly when compared to the likelihood of classical NCs being poorly received by consumers). Let me describe what NCI has done and you judge for yourself.
The Corporate NC remains true to the original notion of a NC, although a little fuzzy about corporate legacy applications. On the desktop side, the Corporate NC probably remains stuck in the first generation clothing consisting of hacked Unix, but implements the JVM, a Netscape Browser, a few other constant program companions, and other goodies like a SmartCard. The server side is pure Unix, supporting Xwindows, terminal emulation, SmartCard desktop indentification, and the ability to serve up Java Applets. When I say pure Unix, I mean something shocking. You can buy a version of the Corporate NC system in which the server side consists of the so-called NCOS, which now stands for Network Computer Server Operating System-and that's exactly what it is, a stand-alone operating system. (You might remember us referring to a NC/OS, but meaning a client-side OS, and hopefully meaning VxWorks.)
If you are wondering why the top brass at NCI recently got fired, now you know. They were building their own server-side OS. Fortunately, since NCOS is Unix-based, it was trivial to offer the server software in application form, which they now do for Sun's Solaris OS. Soon they expect to support most flavors of Unix and, of course, Windows NT. They finally got the server software essentially right, but at what amount of wasted effort?
I don't believe the Corporate NC involves VxWorks. Probably the NC
Desktop has been caste in stone for ages while attention was given to the server side, with no desire to stir up problems by simultaneously attempting a major desktop upgrade.
The Consumer NC got defined on the day NCI merged with Navio, bringing in the Netscape camp of developers. Any notion whatsoever of the NC being a Java machine implementing an alternative PC desktop paradigm with perhaps multimedia jazz was forever squashed. The Consumer NC became Enhanced TV. What is Enhanced TV? Enhanced TV is what the set-top box has evolved into, a hybrid machine that dishes out cable TV with the simultaneous capability to engage the internet with synchronized information about whatever TV channel is tuned in. Great for interactive TV games, sports, etc. Alternatively, the user could simply surf the internet similar to WebTV.
Implementation of Consumer NC has nothing obvious in common with the Corporate NC. Nothing. The server side is based on (surprise!) Netscape's Enterprise software suite. No reinventing server software here.
The client side normally will be literally a set-top box, so it needs to be thought of more as a Web TV rather than a maintenance-free PC. Consumers need have no fear of the client side. It probably looks like a traditional cable box.
The client side software comes in two flavors called a TV Navigator or a DTV Navigator. If you guessed that the D stands for "digital", you guessed right. In either case, the set-top-box software is APPLICATION SOFTWARE ONLY. It is not an integrated OS and application. NCI's stated policy is that they are agnostic about the underlying RTOS needed to run the Navigators. The box manufacturer can put it on any RTOS desired.
Frankly, I have to admit that the Navio troops obviously have their heads on straight. I'm disappointed that the Consumer NC isn't more like the original concept, but that's just my bias. It appears like the Consumer NC is headed for stardom, which is the best proof possible that Navio's vision probably is the most practical.
Suppose you decide to make Consumer NC's based on the NCI Navigator model. What RTOS are you going to use? Before you decide, you should know that Navio developed TV and DTV Navigators using VxWorks. The fact that NCI specifies a real-time operating system suggests that timing might be an issue, and any timing issues obviously already have been worked out with VxWorks (at least on developmental hardware). With VxWorks there will not be any API incompatibilities, or any trouble porting the Netscape browser. Finally, VxWorks provides you with more processor choices than any other RTOS. Unless you are the kind of developer that tweaks operating systems when you should be writing applications, the answer is obvious. And notice that Windows CE would not even be considered, being inadequate and expensive.
For sure the redefined Consumer NC should bode well for WIND. Perhaps downstream the Corporate NC will contribute as well.
Allen
PS - With the proper separation between application and operating system responsibilities, and knowing your organization core capabilities, as illustrated by Navio, it becomes clear why RTOS companies had no business being early investors in Interactive TV or VOD. And this delineation is why WIND could end up dominating the set-top box space without ever lifting a finger to accommodate the vertical market, and with precious few announced set-top box design wins to date. The dominant RTOS will dominate important vertical markets, not the RTOS that embraces characteristics of the vertical market. |
