[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference csc32::consolemanager

Title:POLYCENTER Console Manager
Notice:Kits, Scans, Docs on CSC32:: as PCM$KITS:,PCM$DOCS:, PCM$SCANS:
Moderator:CSC32::BUTTERWORTH
Created:Thu Aug 06 1992
Last Modified:Fri Jun 06 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:1541
Total number of notes:6564

398.0. "PCM cpu guidelines?" by CGOOA::POPOFF () Thu Sep 08 1994 22:30

    Are there any guidelines or 'rules of thumb' that can be used to
    determine what cpu type should be used for PCM?  eg 10 node = 3100/20
    etc.  We have a site that is currently running VCS on 2 4000/90's with 
    about 60 nodes that are starting to crawl.  They are upgrading to PCM.
    Any suggestions?
    
    Thanks
    
    John
T.RTitleUserPersonal
Name		DateLines
398.1OPG::PHILIPAnd through the square window...Fri Sep 09 1994 11:5411
John,

  Nope, we dont have any guidelines other than those posted
  in note 7.1 which relate to memory sizing.

  I should say however, thet your performance should improve
  with PCM, at least, that has been our experience here.

Cheers,
Phil
398.2Not just the CPU chipOPG::SIMONFri Sep 09 1994 12:2316
    John,
         it's also rather simplistic to just look at the raw MIPS of a CPU.
    
    The crawling 400 90's are they really crawling because the CPU is flat
    out? Is the problem that the memory on the system is in short supply or
    an I/O bottleneck to a spindle. VCS always had the problem of a single
    logging spindle, so that is often an area where "crawling" problems are
    seen just because the disk cannot keep up. PCM of course allows you to
    get over that one anyway. I do suggest that you look more closely at
    the systems ( with whatever performance analyzer DEC currently market)
    before looking for guidelines on platform.
    
    AXPs of course fly on CPU intensive bits of PCM (i.e. event scanning),
    but bad I/O problems can still occur with the fastest system.
    
    Cheers Simon....
398.3OSF/1ZENDIA::DBIGELOWInnovate, Integrate, EvaporateFri Sep 09 1994 15:187
    Or go to the OSF/1 (or Ultrix) version. We've had over 200 systems
    (simulated) on 1 host platform giving good results. At one DEC site,
    the entire campus (100+ systems) are being managed by 1 Ultrix based
    platform. You will find that the UNIX based platforms have a much 
    faster IO systems thus improving performance.
    
    Dave