barto@BU-CS.BU.EDU (David Barto) (03/10/88)
>Unix hides the architectural details well enough that I suspect those >who dislike the '86 architecture will give in on this, besides the 386 >helps hide most of the earlier problems at the hardware level. I agree. I just started looking for a system and the 386 based systems are acceptable. However I have yet to find a system which is reasonable in price ($3K) which does not require me to purchase a bound in keyboard and expect a special monitor. For this reason I am still looking. (If anyone finds one which will take a standard CRT as the console, and runs *nix please mail me.) On the mainframe side you mention that you can purchase a small box (or collection of boxes) which will fill the need of large groups without constraining those groups to a single vendor (read IBM or DEC). My father-in-law, president of a medium sized corporation in San Diego, changed his entire computing environment from a few large mainframes to LOTS of Suns and (Celerity) Accels and other reasonable priced processors, with 1 to a few people per machine. The entire computer center quit (or were fired) over the change, and most of the scientists griped that 'they' were losing control of 'their' computing environment. Now, after several months of the new environment, I doubt that they would go back to the centralized approach. I think that those people who care about the future of SCIENTIFIC computing, are aware of the changes and are doing as the above company has done. Those doing the more mundain day-to-day computing of large databases will continue doing it 'the way we have always done it -- The IBM way'. -barto -- David Barto sdcsvax!sdcc6--\ barto@net1.ucsd.edu seismo-!s3sun--!megatek!barto