Path: utzoo!attcan!ncrcan!hcr!chrisp From: chrisp@hcr.UUCP (Chris Phillips) Newsgroups: comp.arch Subject: Re: Memory utilization & inter-process contention Message-ID: <1843@hcr.UUCP> Date: 24 Aug 89 14:40:49 GMT References: <3332@blake.acs.washington.edu> <1989Aug22.163100.25540@utzoo.uucp> <9aid02rf4dNn01@amdahl.uts.amdahl.com> Reply-To: chrisp@spinor.UUCP (Chris Phillips) Organization: HCR Corporation Toronto, Ontario Canada Lines: 29 In article mat@amdahl.UUCP writes: >In article <9aid02rf4dNn01@amdahl.uts.amdahl.com>, sbf10@uts.amdahl.com (Samuel Fuller) writes: >> >> This type of job scheduling is what IBM's MVS operating system does >> [...] >> Is there a happy middle ground? >[...] >allocation. MVS went through an evolution from manual control to >automated control (SRM - System Resource Manager) which Unix has yet to >follow. But even SRM cannot provide a free lunch - it attempts to avoid >overcommitting resources by shutting off or reducing service to some users >according to rules that you provide. Ten years back when I was involved with various parts of MVS(3.6+?) I believe we could show a 30% cpu load on a 43xx class machine with no active work. A good portion of this was the SRM. It (the SRM) was also larger than the UNIX(tm) of that time (1976-1979) (And mostly written in PL/S II). This is not to say that any form of intelligent resource management is bad, but rather to emphasize that it will cost. I think IBM is now providing an AI expert system to generate the parameters for the SRM. This (at least early in MVS life) used to be nearly a Black Art. I can imagine an interface based on NeXTStep would be quite impressive! -- Chris Phillips, Hcr Corp.,1001 "A wise old owl sat in an oak, 130 Bloor St W,Toronto Ontario The more he heard the less he spoke, Bell: (416) 922-1937 M5S 1N5 The less he spoke the more he heard, CIS : 75026,3673 CANADA Why aren't we all like that wise old bird?" UUCP: {utzoo,utcsri,ihnp4}!hcr!chrisp