Hi Guys, I''m playing with Blade 6300 to check performance of compressed ZFS with Oracle database. After some really simple tests I noticed that default (well, not really default, some patches applied, but definitely noone bother to tweak disk subsystem or something else) installation of S10U3 is actually faster than S10U4, and a lot faster. Actually it''s even faster on compressed ZFS with S10U3 than on uncompressed with S10U4. My configuration - default Update 3 LiveUpgraded to Update 4 with ZFS filesystem on dedicated disk, and I''m working with same files which are on same physical cylinders, so it''s not likely a problem with HDD itself. I''m doing as simple as just $time dd if=file.dbf of=/dev/null in few parallel tasks. On Update3 it''s somewhere close to 11m32s and on Update 4 it''s around 12m6s. And it''s both reading from compressed or uncompressed ZFS, numbers a little bit higher with compressed, couple of seconds more, which impressive by itself, but difference is the same, and strangest part is that reading file from compressed ZFS on U3 is faster than reading uncompressed with U4. I''m really surprised by this results, anyone else noticed that ? This message posted from opensolaris.org
On 9/25/07 3:37 AM, "Sergiy Kolodka" <sergiy.kolodka at centrelink.gov.au> wrote:> Hi Guys, > > I''m playing with Blade 6300 to check performance of compressed ZFS with Oracle > database. > After some really simple tests I noticed that default (well, not really > default, some patches applied, but definitely noone bother to tweak disk > subsystem or something else) installation of S10U3 is actually faster than > S10U4, and a lot faster. Actually it''s even faster on compressed ZFS with > S10U3 than on uncompressed with S10U4. > > My configuration - default Update 3 LiveUpgraded to Update 4 with ZFS > filesystem on dedicated disk, and I''m working with same files which are on > same physical cylinders, so it''s not likely a problem with HDD itself. >Did you do a ''zpool upgrade -a''?> I''m doing as simple as just $time dd if=file.dbf of=/dev/null in few parallel > tasks. On Update3 it''s somewhere close to 11m32s and on Update 4 it''s around > 12m6s. And it''s both reading from compressed or uncompressed ZFS, numbers a > little bit higher with compressed, couple of seconds more, which impressive by > itself, but difference is the same, and strangest part is that reading file > from compressed ZFS on U3 is faster than reading uncompressed with U4. > > I''m really surprised by this results, anyone else noticed that ? >I''m running a ''motley group of disks'' on an e450 acting as our jumpstart server and server build times are noticeably quicker since u4.> > This message posted from opensolaris.org > _______________________________________________ > zfs-discuss mailing list > zfs-discuss at opensolaris.org > mail.opensolaris.org/mailman/listinfo/zfs-discuss-Andy --
Just did #zpool upgrade -a, it was already ZFS 4, but anyway. Nothing changed. Any ideas ? This message posted from opensolaris.org