Sorry Jim et al, I had a bunch of replies go directly to Jim and not
the list. I have to remember to Reply All here.
So to recap, the initiator (esx windows 2003 quad cpu guest) showed
consistent cpu usage of 20% throughout the benchmark.
The target (2x quad xeon, 4GB RAM) showed CPU usage from 5-20% during
the benchmark. But during the larger I/O write tests the kernel usage
would shoot to 99% every 5 or so seconds and I/O would stall, this
happened pretty consistently, but more dramatically with the 16K
random writes.
Here are the final numbers:
SQLIO Test
Type:
Drives: d
Test File: sqlio.dat
Test Size: 8000
Threads: 1
Seconds: 300
Outstanding IO: 4
Block Size(s): 4 8 16 32 64
Buffer Setting (N:none, Y:all, S:system, H:hba (Default N)): N
Latency Measurement (S:system, P:processor (Default S)): S
Processor Affinity Mask: 0x0
Iterations: 1
Patterns:
Press any key to continue . . .
Latency
Operation IOs/sec MBs/sec Min/Avg/Max
4K Sequential Read 5925.55 23.14 0/0/452
4K Sequential Write 3446.27 13.46 0/0/549
4K Random Read 603.54 2.35 0/6/269
4K Random Write 558.71 2.18 0/6/2172
8K Sequential Read 3455.82 26.99 0/0/1039
8K Sequential Write 4669.51 36.48 0/0/1904
8K Random Read 646.38 5.04 0/5/176
8K Random Write 4637.31 36.22 0/0/3782
16K Sequential Read 1047.40 16.36 0/3/1782
16K Sequential Write 2509.46 39.21 0/1/9161
16K Random Read 477.66 7.46 0/7/476
16K Random Write 1411.88 22.06 0/2/18710
32K Sequential Read 771.32 24.10 0/4/5447
32K Sequential Write 2100.48 65.64 0/1/3180
32K Random Read 453.89 14.18 0/8/291
32K Random Write 1518.20 47.44 0/2/4298
64K Sequential Read 438.15 27.38 0/8/1355
64K Sequential Write 1095.49 68.46 1/3/2204
64K Random Read 375.51 23.46 0/10/480
64K Random Write 851.65 53.22 1/4/3477
To give an idea, here are numbers from a 6 disk RAID10 on Linux (3
mirrors, so 33% more random IOPS). This is going direct to the
hardware RAID from the kernel iscsi target, no page cache, no read
cache but there is a BBU write-back cache on the controller, hence the
large write numbers.
SQLIO Test
Type:
Drives: k
Test File: sqlio.dat
Test Size: 8000
Threads: 1
Seconds: 300
Outstanding IO: 4
Block Size(s): 4 8 16 32 64
Buffer Setting (N:none, Y:all, S:system, H:hba (Default N)): N
Latency Measurement (S:system, P:processor (Default S)): S
Processor Affinity Mask: 0x0
Iterations: 1
Patterns:
Press any key to continue . . .
Latency
Operation IOs/sec MBs/sec Min/Avg/Max
4K Sequential Read 9573.20 37.39 0/0/605
4K Sequential Write 13608.13 53.15 0/0/661
4K Random Read 739.40 2.88 0/4/737
4K Random Write 1525.71 5.95 0/2/912
8K Sequential Read 7275.31 56.83 0/0/101
8K Sequential Write 9271.87 72.43 0/0/648
8K Random Read 713.16 5.57 0/5/135
8K Random Write 1290.59 10.08 0/2/1249
16K Sequential Read 5361.30 83.77 0/0/213
16K Sequential Write 6390.45 99.85 0/0/1098
16K Random Read 667.07 10.42 0/5/295
16K Random Write 1204.21 18.81 0/2/924
32K Sequential Read 3477.24 108.66 0/1/729
32K Sequential Write 3531.83 110.36 0/0/775
32K Random Read 564.99 17.65 0/6/698
32K Random Write 985.38 30.79 0/3/1111
64K Sequential Read 1744.84 109.05 1/2/753
64K Sequential Write 1776.51 111.03 0/1/47
64K Random Read 522.74 32.67 0/7/212
64K Random Write 756.41 47.27 0/4/1189
Is there anything I can try on my end to bring iscsitgt read numbers
more in line with the Linux target's?
-Ross
_______________________________________________
storage-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/storage-discuss
