On Aug 15, 2009, at 8:34 PM, en...@businessgrade.com wrote:
Quoting Ross Walker <rswwal...@gmail.com>:
On Aug 15, 2009, at 4:46 PM, Mike La Spina <no-
re...@opensolaris.org> wrote:
Hi,
The method you have chosen will not give you accurate read
performance data. The reads will be stored in cache and unless
the iostat call was invoked during a read cycle from disk you
would not see the actual io. Try installing and using bonnie++.
While true it will run from cache the fact that it's 16GB in size
should negate that given the system has 8GB of memory, as long as the
OP runs it twice for each and takes the second value.
But it still only measures asynchronous io performance. You'll need a
benchmark package to see synchronous io performance since dd doesn't
support the 'direct' flag.
-Ross
Thanks for the input everyone. I understand that I should be testing
with a benchmark program. But I'm confused a wee bit I guess because
I would imagine my asynchronous read performance to be much better
than 48MB per second. Right/Wrong? Even if the read performance is
not accurate, it should still be ball park or greater than
synchronous io.
Yes, it should be.
Maybe the controller is in IDE compatibility mode?
-Ross
_______________________________________________
storage-discuss mailing list
storage-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/storage-discuss
