gamiddleton@thunder.UUCP (Guy Middleton) (11/14/85)
We have a very-heavily-loaded Vax here, and I am trying to figure out ways of making it more usable, short of buying another one. One idea is to rewrite the scheduler, so that time is allocated on a per-uid, instead of a per-process basis. Has anybody done this to the kernel? -- -Guy Middleton, Lakehead University, Thunder Bay, Ont. ..{watmath,ihnp4!mmm}!thunder!gamiddleton
david@rsch.wisc.edu (David Parter) (11/15/85)
> We have a very-heavily-loaded Vax here, and I am trying to figure out ways > of making it more usable, short of buying another one. One idea is to > rewrite the scheduler, so that time is allocated on a per-uid, instead of a > per-process basis. Has anybody done this to the kernel? see The Fair Share Scheduler G. J. Henry in Bell Labs Technical Journal vol 63, october 1984, number 8, part2. when the BLTJ came out, there was much discussion of the price (originally outragously high, then lowered), and how to order a single copy from ATT, but i do not remember now. david -- david parter UWisc Systems Lab uucp: ...!{allegra,harvard,ihnp4,seismo, topaz}!uwvax!david arpa: david@rsch.wisc.edu
paul@vcvax1.UUCP (paul) (11/15/85)
> We have a very-heavily-loaded Vax here, and I am trying to figure out ways > of making it more usable, short of buying another one. One idea is to > rewrite the scheduler, so that time is allocated on a per-uid, instead of a > per-process basis. Has anybody done this to the kernel? See "The Fair Share Scheduler" by G.J. Henry in the last "UNIX System" issue of the the Bell Labs Technical Journal (vol 63. No.8 Part 2, October 1984). I suspect, however, that the scheduler described there is only available within Bell Labs. Paul Kleppner VenturCom, Inc. {harvard,mit-eddie}!cybvax0!vcvax1!venix