On Sat, 17 Jun 2000, David Murphy wrote:
> If you only ever run one process at a time, definitely, otherwise, we
> don't yet know who they compare.
????
what does single/multiple processes have to do with it? the raid
card/software has no concept of processes.
if you mean low/high I/O pressure: again that's irrelevant! at this
this low level the only metric is bit rate - and RAID0 is king,
followed by RAID5 (for reads). RAID1 is last.
if you mean the pattern of I/O (random vs seq'tial), bit-rate is
still king. Even though RAID1 per block seek time is lower ( =<
highest seek time of component disks + disk read) than RAID5. (=
highest seek time + 3 disk read[1]), over a collection of random disk
blocks RAID5 should still win because of it's greater bit-rate.
that is what i have seen. You argue there must be a deficiency
therefore in linux RAID1[2]. However i (and others) still feel that
RAID5 as a technology /must/ be faster than RAID1.
Either accept the weight of opinion and let the thread die, or come
up with numbers.
regards,
--
Paul Jakma paul at clubi.ie
PGP5 key: http://www.clubi.ie/jakma/publickey.txt
-------------------------------------------
Fortune:
In defeat, unbeatable; in victory, unbearable.
-- W. Churchill, on General Montgomery
[1]. Although this could be done in parallel with multiple
controllers.
[2]. you cited round-robin reads a la Sun. However linux does do
this.
Maintained by the ILUG website team. The aim of Linux.ie is to
support and help commercial and private users of Linux in Ireland. You can
display ILUG news in your own webpages, read backend
information to find out how. Networking services kindly provided by HEAnet, server kindly donated by
Dell. Linux is a trademark of Linus Torvalds,
used with permission. No penguins were harmed in the production or maintenance
of this highly praised website. Looking for the
Indian Linux Users' Group? Try here. If you've read all this and aren't a lawyer: you should be!