blisenko - I see it about the way you do. Linux is making strong inroads on the low end of the market, especially web farms, and is beginning to also capture some file and print, especially small and medium business. It is the best product for that market as far as I can tell - clean, efficient, robust, cheap... I see Linux expanding horizontally to more of that class of apps - relatively simple function, controlled configurations where the limited support is not a problem because a Unix-savvy admin can do it all before his first coffee break.
NT - Win2K - will push up somewhat into the mid-tier applications. It is hampered by a poor management infrastructure which is only partially addressed by the latest product, and which will require enhancements by the OEM partners and middleware vendors (or by significant enhancements to MSFT's own middleware) which will probably take years to achieve. Likewise, hardware advances will allow NT to grow onto larger hardware platforms like the Unisys 32-way, but that just opens the door to development which will effectively use that hardware - again, a process which will take years.
The notion that NT will "displace" Unix is, in my mind, silly - there is no dynamic which would drive that. There was not sufficient pressure to even converge the various high end Unix variants... Solaris will almost certainly be the defacto standard for high end Unix but HP-UX won't go away any time soon. AIX is a little weaker only because IBM has never been a whole-hearted Unix supporter - they make a lot more from MVS and OS400 - and as IBM shifts away from the platform business - Gerstner's "dumbell" - AIX may get replaced by somebody else's Unix, or maybe by Monterray... CPQ's DEC Unix or Tru64 or whatever they call it today is a technically good product -superior even, both in performance and ruggedness - but it is not from the same heritage as Solaris or even HP-UX. It came out of the Mach kernel effort at Carnegie-Mellon (Gordon Bell and Bill Strecker were deeply involved at CMU for much of their careers) and then got further tainted with the abortive OSF effort. I think that product will also eventually be replaced by "someone else's Unix".
Solaris is a contender to be that replacement, except that SUNW has not shown much interest in moving down that path.
In any event, the steady state model has proprietary stuff in the very high end - Tandem NSK on Himalaya, MVS or other mainframe variants... but increasingly, Unix will pressure the bottom of even that market as the reliability and dynamic configurability improve. Solaris is the likely big winner, with HP-UX remaining as an important force. Unix in turn will be pressured on the low end by NT, but still the overall Unix market will continue to grow. NT will be pressured in it's low end and some horizontal markets by Linux. There is no force I can see which would kill any of these products, or which would allow them to dominate greatly outside of their target markets...
Your comments about not needing an OS may apply more and more to things individuals use - it is already true of many "internet appliances" such as the folks are using in Japan and Finland... but traditional desktops will also continue to evolve, although much less rapidly in the past...
So much for my crystal ball, that's how I see it... |
