More on the load stuff at tdi2

[allbery at ncoast.UUCP]

Well, it seems I forgot some information:

(1) We've already fiddled with buffers extensively, to no avail.
(2) With 5MB we don't swap.  With 8MB we don't swap.  And we're still slow.
(3) I've done more testing; the only correlation between system activity and
    slowness is on number of processes.

We usually run 1 user in an Informix-SQL application, 7 in an RM/COBOL
application, and 5 in a UNIFY application; the others do various random
stuff (1 spreadsheet, 2 vi, ...)

ho95e!wcs (Bill Stewart) suggests we upgrade to a paging system.  By (2)
above I suspect it wouldn't help, but paging OSes don't seem to be available
for the Plexus.  (I even asked mtXinu if they knew of a BSD4.2 for it...)

BTW, the IMSP disk controller is too dumb for two disks to reduce disk-bound
processing time; it treats the two disks as one large disk.  However, lots
of disk-bound stuff takes place at night, and runs blindingly fast.  Once a
week the nightly processing overlaps for many software systems, but it still
runs quickly.  It's just daytime stuff that drags.
-- 
++Brandon (Resident Elf @ ncoast.UUCP)
 ____   ______________
/    \ / __   __   __ \   Brandon S. Allbery	    <backbone>!ncoast!allbery
 ___  | /__> /  \ /  \    aXcess Co., Consulting    ncoast!allbery at Case.CSNET
/   \ | |    `--, `--,    6615 Center St. #A1-105 	   (... at relay.CS.NET)
|     | \__/ \__/ \__/    Mentor, OH 44060-4101     
\____/ \______________/   +1 216 974 9210