"Try an SGA more like 20-25 GB. Remember, the database can cache more
effectively than any file system underneath. The best I/O is the I/O
you don't have to make."
We'll be turning up the SGA size from 4GB to 16GB.
The arc size will be set from 8GB to 4GB.
"This can be a red herring. Judging by the number of IOPS below,
it has not improved. At this point, I will assume you are using
disks that have NCQ or CTQ (eg most SATA and all FC/SAS drives).
If you only issue one command at a time, you effectively disable
NCQ and thus cannot take advantage of its efficiencies."
Here's another sample of the data taken at another time after the number of
concurrent ios change from 10 to 1. We're using Seagate Savio 10K SAS
drives...I could not pull up info if the drives support NCQ or not. What's the
recommended value to set concurrent IOs to?
r/s w/s Mr/s Mw/s wait actv wsvc_t asvc_t %w %b device
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 c0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 c0t0d0
1402.2 7805.3 2.7 36.2 0.2 54.9 0.0 6.0 0 940 c1
10.8 1.0 0.1 0.0 0.0 0.1 0.0 7.0 0 7 c1t0d0
117.1 640.7 0.2 1.8 0.0 4.5 0.0 5.9 1 76 c1t1d0
116.9 638.2 0.2 1.7 0.0 4.6 0.0 6.1 1 78 c1t2d0
116.4 639.1 0.2 1.8 0.0 4.6 0.0 6.0 1 78 c1t3d0
116.6 638.1 0.2 1.7 0.0 4.6 0.0 6.1 1 77 c1t4d0
113.2 638.0 0.2 1.8 0.0 4.6 0.0 6.1 1 77 c1t5d0
116.6 635.3 0.2 1.7 0.0 4.5 0.0 6.0 1 76 c1t6d0
116.2 637.8 0.2 1.8 0.0 4.7 0.0 6.2 1 79 c1t7d0
115.3 636.7 0.2 1.8 0.0 4.4 0.0 5.8 1 77 c1t8d0
115.4 637.8 0.2 1.8 0.0 4.5 0.0 5.9 1 77 c1t9d0
114.8 635.0 0.2 1.8 0.0 4.3 0.0 5.7 1 76 c1t10d0
114.9 639.9 0.2 1.8 0.0 4.7 0.0 6.2 1 78 c1t11d0
115.1 638.7 0.2 1.8 0.0 4.4 0.0 5.9 1 77 c1t12d0
1.6 140.0 0.0 15.1 0.0 0.6 0.0 4.4 0 8 c1t13d0
1.3 9.1 0.0 0.1 0.0 0.0 0.0 1.0 0 0 c1t14d0
--
This message posted from opensolaris.org
_______________________________________________
zfs-discuss mailing list
zfs-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
