On Sat, 1 Aug 2009, Louis-Frédéric Feuillette wrote:
I find the results suspect. 1.2GB/s read, and 500MB/s write ! These are
impressive numbers indeed. I then looked at the file sizes that iozone
used... How much memory do you have? I seems like the files would be
able to comfortably fit in memory. I think this test needs to be re-run
with Large files (ie >2*Memory size ) for them to give more accurate
data.
The numbers are indeed "suspect" but the iozone sweep test is quite
useful in order to see the influence of zfs's caching via the ARC.
The sweep should definitely be run to at least 2X the memory size.
Unrelated, what did you use to generate those graphs, they look good.
Iozone output may be plotted via gnuplot or Microsoft Excel. This
looks like the gnuplot output.
Bob
--
Bob Friesenhahn
bfrie...@simple.dallas.tx.us, http://www.simplesystems.org/users/bfriesen/
GraphicsMagick Maintainer, http://www.GraphicsMagick.org/
_______________________________________________
zfs-discuss mailing list
zfs-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
