Hi,
Panic over I think. It was more the weakness of my testing rather than any
real failing on the setup.
tar may impose certain overheads which make it a less than efficient tool
for testing this kind of thing.
Thanks to Alan for his feedback.
--------------------------------originally
--------------------------------------------
Just been doing a couple of very crude tests on disk throughput. System is
an ES40 4x500, 4Gb RAM. There's an RA3000 with 12x36GB drives, 64Mb cache on
the controller, which houses 2 RAID5 volumes of about 150Gb each using
ADVFS. There's also another 150Gb volume on the internal controller which is
an ADVFS volume striped across 4x 36GB drives. If I do a "tar -cf /dev/null"
on one of the volumes on the RA iostat on that "disk" shows a transfer of
around 4.5Mb/s in iostat. If I do the same on the internal disks I see
numbers up to 12Mb/s.
I can't help thinking the volumes on the RA ought to be a bit quicker. I
hasten to add I know this is a seriously crude test and it's hardly
scientific but there's still that nagging doubt.
Is it possible I've missed a setting on the RA or am I just being hatstand
cos it's Friday night?
Cheers
Joe
Disclaimer: Sorry for the disclaimer.
MetaPack
The Lightwell
12/16 Laystall Street
Clerkenwell
London EC1R 4PF
Tel: +44 (0) 20 7843 6720
Fax: +44 (0) 20 7843 6721
--------------------------------------------------------------------------
This email is confidential and proprietary;
all information contained in it must be used only by the addressee in
accordance with MetaPack's terms of business and non-disclosure agreement.
Disclosure, copying, and distribution to, or use by, anyone other than the
intended recipient is strictly prohibited and may be unlawful.
Received on Tue Nov 27 2001 - 14:52:53 NZDT