pesv@enea.se (Peter Svenson) (02/23/88)
Maybe there is a more apt newsgroup to post this message in, but here goes anyway.. I wonder whether there is any research going on in the area of advanced O/S designs? What is advanced? Well, like ''the next generation'' I have been wondering if there is no better way to organize a system's structure than the now quite dated directory-subdirectory filesystems. Also most O/S's, no matter how distinct ususally behave in the same general manner when it comes to multitasking, virtual memory, etc. I do not say that this is 'Bad' in any way, I just wonder if there are any fresh ideas that may yet have to be tested due to the current wimpy power of todays computers. OK, ok, I give up! What it is, is, I've read William Gibson's book Neuromancer. If you've read it, you know what I'm up to allready. The current state of the art O/S's work pretty neat.. up to a point. But what about the research in object-oriented O/S's with underlying MASSIVELY parallel architectures, that can support hundred of TERAbytes of *primary* memory, or where the primary and secundary memory blend together. Where all machines on a network (and I'm talking networks over countries here) blend into one single 'entity'. What kind of tricks will an O/S have to do support these kind of things in a coherent manner? I am fully aware of that I am now raving madly, but if (and only if) there is anyone engaged in this kind of work (if you at all can understand what I'm getting at) or have any hints, I'd appreciate all information whatsoever. Thanks in advance, -- Peter (turbo) Svenson pesv@enea (UUCP) enea!pesv@seismo.arpa (ARPA) -------------------------------------------------------------------------------- The Kabbalistic Karate Viking from outer space.Save it, chum. Long time, no Foo
thompson@calgary.UUCP (Bruce Thompson) (02/27/88)
In article <2760@enea.se>, pesv@enea.se (Peter Svenson) writes: > > I wonder whether there is any research going on in the area of advanced > O/S designs? What is advanced? Well, like ''the next generation'' > I have been wondering if there is no better way to organize a system's > structure than the now quite dated directory-subdirectory filesystems. > Also most O/S's, no matter how distinct ususally behave in the same > general manner when it comes to multitasking, virtual memory, etc. > > Thanks in advance, > -- > Peter (turbo) Svenson pesv@enea (UUCP) enea!pesv@seismo.arpa (ARPA) > -------------------------------------------------------------------------------- > The Kabbalistic Karate Viking from outer space.Save it, chum. Long time, no Foo Just to throw my 2 cents worth in, I would like to point people in the direction of Honeywell's MULTICS. This was a machine which was vastly ahead of it's time, and who's main problem now is that it's old, and the hardware isn't very fast. We have a large installation of MULTICS here at the University of Calgary, and at the least, it's impressive. Particularly at the level of virtual memory, and run-time dynamic linking. What is truly unfortunate, is that few operating systems in the big wide world seem to have made use of any of the lessons learned with MULTICS. In these days of HUGE virtual address spaces available to processes, and low-cost high-speed paralel machines (for example the Transputer) becoming available, a MULTICS-like O/S Architecture would seem to fit the bill rather nicely. BTW, MULTICS does use multiple CPUs, but what the organization is, I don't know. TTFN, Bruce Thompson ------------------------------------------------------------------------------ Bruce Thompson | Hey! What're you doing at my University of Calgary, | terminal! (Skuffle skuffle) Computer Science Department | What is this chicken scratch? (403)220-3538 or (403)220-5019 (office) | ^X-^C | ...!alberta!calgary!vaxb!thompson |