on X86, test disk IO with dd on the internal disk with SVM RAID-1, the cmd is
# iostat -xcnzC 1
cpu
us sy wt id
7 4 0 89
extended device statistics
r/s w/s kr/s kw/s wait actv wsvc_t asvc_t %w %b device
0.0 1996.8 0.0 105586.4 0.5 210.0 0.3 105.2 0 194 c3
0.0 10.0 0.0 5.5 0.0 0.0 0.0 3.6 0 3 c3t0d0
0.0 995.9 0.0 52681.2 0.5 111.4 0.5 111.8 4 100 c3t2d0
0.0 990.9 0.0 52899.7 0.0 98.6 0.0 99.5 1 90 c3t3d0
0.0 12.0 0.0 188.0 0.0 0.0 0.0 0.2 0 0 c4
0.0 12.0 0.0 188.0 0.0 0.0 0.0 0.2 0 0 c4t0d0
0.0 12.0 0.0 188.0 0.0 0.0 0.0 0.2 0 0 c5
0.0 12.0 0.0 188.0 0.0 0.0 0.0 0.2 0 0 c5t0d0
0.0 11.0 0.0 216.0 0.0 1.7 0.0 158.5 0 68 md/d31
0.0 110.0 0.0 52347.7 0.8 12.7 7.2 115.5 80 92 md/d33
0.0 11.0 0.0 216.0 0.0 1.7 0.0 158.5 0 68 md/d41
0.0 110.0 0.0 52347.7 0.0 12.7 0.0 115.5 0 92 md/d43
0.0 11.0 0.0 216.0 0.0 1.6 0.0 144.3 0 63 md/d51
0.0 111.0 0.0 53243.6 0.0 11.4 0.0 102.5 0 87 md/d53
0.0 12.0 0.0 188.0 0.1 0.0 5.4 0.3 7 0 md/d60
0.0 12.0 0.0 188.0 0.0 0.0 0.0 0.2 0 0 md/d70
0.0 12.0 0.0 188.0 0.0 0.0 0.0 0.2 0 0 md/d80
# metastat -p
d60 -m d70 d80 1
d70 4 1 c4t0d0s0 \
1 c4t1d0s0 \
1 c4t2d0s0 \
1 c4t3d0s0
d80 4 1 c5t0d0s0 \
1 c5t1d0s0 \
1 c5t2d0s0 \
1 c5t3d0s0
d34 -m d44 d54 1
d44 1 1 c3t2d0s4
d54 1 1 c3t3d0s4
d33 -m d43 d53 1
d43 1 1 c3t2d0s3
d53 1 1 c3t3d0s3
d31 -m d41 d51 1
d41 1 1 c3t2d0s1
d51 1 1 c3t3d0s1
d30 -m d40 d50 1
d40 1 1 c3t2d0s0
d50 1 1 c3t3d0s0
d6 -m d16 d26 1
d16 1 1 c3t0d0s6
d26 1 1 c3t1d0s6
d1 -m d11 d21 1
d11 1 1 c3t0d0s1
d21 1 1 c3t1d0s1
d0 -m d10 d20 1
d10 1 1 c3t0d0s0
d20 1 1 c3t1d0s0
look at the controller 3:
The %b of controller 3 is 194% which comes from the addition of all devices
under this controller. is this correct to simply add them up ?
second, the w/s of c3t2d0 & c3t3d0 is over 900 each, but if looking at teh
breakdown of the metadevices, c3t2d0 has d40, d41, d43, d44 with total of
121w/s, c3t3d0 has d50, d51, d53, d54 with total of 122w/s, where are the rest
of 870 IOPS come from ?
I need help on this.
Regards,
Dayu Yang
--
This message posted from opensolaris.org
_______________________________________________
perf-discuss mailing list
[email protected]