We've got an AlphaServer 4100 with 4 CPUs and 4 gigs of memory running
DU4.0D which has been configured by a previous system administrator with
the following swap partitions:
rz19b: 0.9 gig
rz18b: 1 gig
rz17b: 1 gig
rz20b: 1 gig
rz21b: 1 gig
rz22b: 1 gig
rz22h: 3 gig
The disks are all mostly similar:
rz17 at scsi2 target 1 lun 0 (LID=3) (DEC RZ1CB-CA (C) DEC LYJ0) (Wide16)
rz18 at scsi2 target 2 lun 0 (LID=4) (DEC RZ1CB-CA (C) DEC LYJ0) (Wide16)
rz19 at scsi2 target 3 lun 0 (LID=5) (DEC RZ1BB-CS (C) DEC 0656) (Wide16)
rz20 at scsi2 target 4 lun 0 (LID=6) (DEC RZ1CB-CA (C) DEC LYJ0) (Wide16)
rz21 at scsi2 target 5 lun 0 (LID=7) (DEC RZ1CB-CA (C) DEC LYJ0) (Wide16)
rz22 at scsi2 target 6 lun 0 (LID=8) (DEC RZ1CB-CA (C) DEC LYJ0) (Wide16)
And scsi2 is an internal bus on the 4100.
My questions is just what would be the optimal swap setup? Is this setup
overkill? I should probably mention that this machine does do a lot of
number crunching sometimes on datasets which can be very large (100's of
megs). I don't think that this the size of this swap setup is necessary,
but the previous system administrator was concerned about speed, and
therefore split the swap partitions across multiple drives like this.
Will you actually get speed gains out of this many multiple swap paritions
under DU4.0D? And we've been able to reclaim the other internal scsi bus,
so should i move half of the disks to the other bus? Should I add even
more swap partitions? Is that rz22h swap partition doing anything good at
all?
Enquiring minds want to know...
--
Lamont Granquist lamontg_at_raven.genome.washington.edu
Dept. of Molecular Biotechnology (206)616-5735 fax: (206)685-7344
Box 352145 / University of Washington / Seattle, WA 98195
PGP pubkey: finger lamontg_at_raven.genome.washington.edu | pgp -fka
Received on Fri Jan 15 1999 - 08:20:05 NZDT