ken@cs.rochester.edu (Ken Yap) (04/19/88)
Hmm. Unix system binaries don't change that often. What if most of /bin, /usr, /lib were put on a CD? Attach one RO optical drive to your network of workstations. Heck, if the drives are cheap, even one per workstation. A cache would speed access to the most often requested files. Maybe some scheme to allow bypassing distribution binaries with local versions. OS upgrade would be just a matter of sending out CDs. Has this been done already? Or planned? Or too impractical yet/ever? Ken
verber@apatosaur.cis.ohio-state.edu (Mark Verber) (04/19/88)
In article <8786@sol.ARPA> ken@cs.rochester.edu (Ken Yap) writes: >Hmm. Unix system binaries don't change that often. What if most of >/bin, /usr, /lib were put on a CD? Attach one RO optical drive to your >network of workstations. Heck, if the drives are cheap, even one per >workstation. A cache would speed access to the most often requested >files. Maybe some scheme to allow bypassing distribution binaries with >local versions. OS upgrade would be just a matter of sending out CDs. > >Has this been done already? Or planned? Or too impractical yet/ever? > > Ken The optical disks are still too slow for this to be practical. If someone wanted to do this a good starting place would be the work done a BRL which permitted the root file system to be on write-protected disks. On the other hand dist. for sources would be ideal on CDs. Something I hope people start doing. I believe that DEC is doing this for VMS stuff now. Cheers, Mark A. Verber
greg@ncr-sd.SanDiego.NCR.COM (Greg Noel) (04/21/88)
In article <8786@sol.ARPA> ken@cs.rochester.edu (Ken Yap) writes: >Hmm. Unix system binaries don't change that often. What if most of >/bin, /usr, /lib were put on a CD? .... >Has this been done already? Or planned? Or too impractical yet/ever? This was discussed a couple of months ago, in the context of various plans for read-only file systems. In article <10986@tut.cis.ohio-state.edu> verber@apatosaur.cis.ohio-state.edu (Mark Verber) writes: >The optical disks are still too slow for this to be practical. If you insist that the file system be a "traditional" Unix file system, then I'd have to agree with you. But if you give me license to create a file system with Unix semantics (think of accessing it via the file system switch), then I'm not so sure. With some careful advance planning (remember, you can afford to do a lot of planning, since you only write the disk once), I think it should be possible to get to any file with a single seek. This would make the access fast enough for a single-user workstation, for example -- a half-second or so to load a program. >If someone wanted to do this a good starting place would be the work done >a BRL which permitted the root file system to be on write-protected disks. BRL? Do you mean the AFHQSC at the Pentagon? (If so, that's pretty close; they're only a few miles apart....) BRL may have done it, as well, but I think the first folks to do it were the Air Force. -- -- Greg Noel, NCR Rancho Bernardo Greg.Noel@SanDiego.NCR.COM or greg@ncr-sd
davidsen@steinmetz.ge.com (William E. Davidsen Jr) (04/21/88)
In article <10986@tut.cis.ohio-state.edu> verber@apatosaur.cis.ohio-state.edu (Mark Verber) writes: > ... >On the other hand dist. for sources would be ideal on CDs. Something >I hope people start doing. I believe that DEC is doing this for VMS >stuff now. Sources would be nice, how about the man pages. Usually the man access is slow enough that an extra 250ms wouldn't hurt anything. -- bill davidsen (wedu@ge-crd.arpa) {uunet | philabs | seismo}!steinmetz!crdos1!davidsen "Stupidity, like virtue, is its own reward" -me
dan@rna.UUCP (Dan Ts'o) (04/24/88)
In article <2177@ncr-sd.SanDiego.NCR.COM> greg@ncr-sd.SanDiego.NCR.COM (Greg Noel) writes: >In article <8786@sol.ARPA> ken@cs.rochester.edu (Ken Yap) writes: >>Hmm. Unix system binaries don't change that often. What if most of >>/bin, /usr, /lib were put on a CD? .... >>Has this been done already? Or planned? Or too impractical yet/ever? > >This was discussed a couple of months ago, in the context of various >plans for read-only file systems. We're planning to investigate WORM drives for archiving data. Can someone summarize the state of UNIX software for WORMs ? Thanks. Cheers, Dan Ts'o 212-570-7671 Dept. Neurobiology dan@rna.rockefeller.edu Rockefeller Univ. ...cmcl2!rna!dan 1230 York Ave. rna!dan@nyu.edu NY, NY 10021 tso@rockefeller.edu