From:	SMTP%"RELAY-INFO-VAX@CRVAX.SRI.COM" 16-SEP-1993 08:32:50.13
To:	EVERHART
CC:	
Subj:	Re: VAXcluster reconfiguration question

X-Newsgroups: comp.os.vms
From: fairfield@slacvx.slac.stanford.edu
Subject: Re: VAXcluster reconfiguration question
Message-Id: <1993Sep14.155725.1@slacvx.slac.stanford.edu>
Lines: 37
Sender: news@unixhub.SLAC.Stanford.EDU
Organization: Stanford Linear Accelerator Center
Date: Tue, 14 Sep 1993 23:57:25 GMT
To: Info-VAX@kl.sri.com
X-Gateway-Source-Info: USENET

In article <15595893@MVB.SAIC.COM>, system@meng.ucl.ac.uk (Mark Iline) writes:
> 
> It's already been stated that having a local page/swap disk is a good idea, 
> to avoid paging or swapping over the network, and the accompanying network 
> load and overhead. This is good advice.
> 
> However, you should bear in mind that a significant amount of paging can 
> occur from the system disk, as this is where many of the commonly used 
> executables (eg COPY), and the various sharable RTLs reside. Having these 
> images installed /SHARED can significantly reduce this paging, but the 
> effectiveness of this is very dependant on workload. (eg on a workstation, 
> how likely is it that two processes will be running COPY simultaneously ?).
[...] 
> Another possible method is to redefine SYS$SYSROOT so that executables are 
> found on a local disk before the system disk. This needs to be done early 
> in the start-up procedure - ie before the images are INSTALled.

    I've been dusing the method described in the last paragraph since VMS V5.0.
It works very well indeed.  You don't need to copy _all_ the system images,
just the ones you think will have a high likelihood of getting a "hit".

    I found that SATELLITE_PAGE.COM was an appropriate place to redefine
SYS$SYSROOT: the device/directory that you want to stick into SYS$SYSROOT
is sure to be there (if the MOUNT succeeds) and it early enough in the
boot that the INSTALL (or most of them) haven't yet been done.  Finally,
I suggest you put the local disk _between_ SYS$SPECIFIC and SYS$COMMON:
it can be very frustrating to try to look at what should have been a log
file in SYS$SPECIFIC that actually got written to the local disk...and
you can't mount the local disk, etc.

	-Ken
-- 
 Dr. Kenneth H. Fairfield    |   Internet: Fairfield@Slac.Stanford.Edu
 SLAC, P.O.Box 4349, MS 98   |   DECnet:   45537::FAIRFIELD (45537=SLACVX)
 Stanford, CA   94309        |   FAX:      (415) 926-4335
 ----------------------------------------------------------------------------
 These opinions are mine, not SLAC's, Stanford's, nor the DOE's...