On 04/08/2014 05:44 PM, Mike Snitzer wrote:
How is it that you've extracted these results from the test run? I'd be
open to rerunning the same on my testbed (1 Intel Xeon E5649 Westmere-EP
with 12 logical cores, AES, and 6 SAS devices for RAID, 1 STEC PCIe SSD)
I could also yield this testbed to you if that'd be easier than you
telling me how to extract the test results.
Thanks Mike!
Pasting new results for loaned machine and things are getting more
interesting (with respect to Marian's reply):
single (rotational) device:
TEST NAME OPERATION AGGRB_A KB/s AGGRB_B KB/s DIFF A->B
------------------------- --------- ------------ ------------ ---------
disk_nocrypt-read READ 479129 478528 -0.13 %
disk_nocrypt-write WRITE 225040 225718 0.30 %
disk_nocrypt-randread READ 30150 30193 0.14 %
disk_nocrypt-randwrite WRITE 39059 39546 1.25 %
disk_nocrypt-rw READ 92477 94021 1.67 %
disk_nocrypt-rw WRITE 93242 94800 1.67 %
disk_nocrypt-randrw READ 14426 14902 3.30 %
disk_nocrypt-randrw WRITE 14388 14862 3.29 %
zero-read READ 272889 354069 29.75 %
zero-write WRITE 145721 306175 110.11 %
zero-randread READ 482888 599366 24.12 %
zero-randwrite WRITE 151175 346728 129.36 %
zero-rw READ 91280 131950 44.56 %
zero-rw WRITE 92036 133042 44.55 %
zero-randrw READ 72625 142476 96.18 %
zero-randrw WRITE 72433 142099 96.18 %
disk-read READ 193161 210156 8.80 %
disk-write WRITE 33845 182790 440.08 %
disk-randread READ 29746 29640 -0.36 %
disk-randwrite WRITE 7783 31689 307.16 %
disk-rw READ 23954 69686 190.92 %
disk-rw WRITE 24152 70263 190.92 %
disk-randrw READ 6770 12523 84.98 %
disk-randrw WRITE 6753 12490 84.95 %
single PCIe SSD device:
(Staring to immense difference between sequential reads and random
reads. Can it be there some sort of raid inside SSD card?)
TEST NAME OPERATION AGGRB_A KB/s AGGRB_B KB/s DIFF A->B
------------------------- --------- ------------ ------------ ---------
disk_nocrypt-read READ 1072537 1145446 6.80 %
disk_nocrypt-write WRITE 843924 828586 -1.82 %
disk_nocrypt-randread READ 2091110 2108825 0.85 %
disk_nocrypt-randwrite WRITE 1149542 1137971 -1.01 %
disk_nocrypt-rw READ 425704 426661 0.22 %
disk_nocrypt-rw WRITE 429228 430193 0.22 %
disk_nocrypt-randrw READ 534206 532311 -0.35 %
disk_nocrypt-randrw WRITE 532791 530900 -0.35 %
zero-read READ 271388 351871 29.66 %
zero-write WRITE 143748 300000 108.70 %
zero-randread READ 487264 600310 23.20 %
zero-randwrite WRITE 148958 344308 131.14 %
zero-rw READ 90927 131659 44.80 %
zero-rw WRITE 91679 132748 44.80 %
zero-randrw READ 71325 137468 92.73 %
zero-randrw WRITE 71136 137104 92.74 %
disk-read READ 243854 269678 10.59 %
disk-write WRITE 179773 300408 67.10 %
disk-randread READ 518848 647679 24.83 %
disk-randwrite WRITE 195314 348398 78.38 %
disk-rw READ 103381 131270 26.98 %
disk-rw WRITE 104237 132356 26.98 %
disk-randrw READ 93681 145719 55.55 %
disk-randrw WRITE 93433 145333 55.55 %
4 rotational devices forming md raid5, 64KiB chunk size:
TEST NAME OPERATION AGGRB_A KB/s AGGRB_B KB/s DIFF A->B
------------------------- --------- ------------ ------------ ---------
disk_nocrypt-read READ 560436 547773 -2.26 %
disk_nocrypt-write WRITE 360552 359039 -0.42 %
disk_nocrypt-randread READ 62256 62118 -0.22 %
disk_nocrypt-randwrite WRITE 51591 52289 1.35 %
disk_nocrypt-rw READ 59680 58955 -1.21 %
disk_nocrypt-rw WRITE 60174 59443 -1.21 %
disk_nocrypt-randrw READ 18328 18219 -0.59 %
disk_nocrypt-randrw WRITE 18280 18171 -0.60 %
zero-read READ 274712 353651 28.74 %
zero-write WRITE 146469 300904 105.44 %
zero-randread READ 498736 603246 20.95 %
zero-randwrite WRITE 150131 355003 136.46 %
zero-rw READ 92481 132359 43.12 %
zero-rw WRITE 93246 133455 43.12 %
zero-randrw READ 72278 141164 95.31 %
zero-randrw WRITE 72086 140790 95.31 %
disk-read READ 177154 203054 14.62 %
disk-write WRITE 71057 241913 240.45 %
disk-randread READ 58757 58988 0.39 %
disk-randwrite WRITE 14453 28966 100.42 %
disk-rw READ 59772 67560 13.03 %
disk-rw WRITE 60267 68120 13.03 %
disk-randrw READ 9634 12815 33.02 %
disk-randrw WRITE 9608 12781 33.02 %
4 rotational devices forming md raid5, 512KiB chunk size:
(I'll repeat this one since there's strange increase in random read
performance on raw raid5 device, see line "disk_nocrypt-randread"):
TEST NAME OPERATION AGGRB_A KB/s AGGRB_B KB/s DIFF A->B
------------------------- --------- ------------ ------------ ---------
disk_nocrypt-read READ 734811 730206 -0.63 %
disk_nocrypt-write WRITE 264474 259773 -1.78 %
disk_nocrypt-randread READ 64387 83218 29.25 %
disk_nocrypt-randwrite WRITE 52671 51898 -1.47 %
disk_nocrypt-rw READ 75785 73469 -3.06 %
disk_nocrypt-rw WRITE 76413 74077 -3.06 %
disk_nocrypt-randrw READ 17664 17690 0.15 %
disk_nocrypt-randrw WRITE 17617 17643 0.15 %
zero-read READ 270967 354099 30.68 %
zero-write WRITE 145277 307230 111.48 %
zero-randread READ 478481 585805 22.43 %
zero-randwrite WRITE 149734 349997 133.75 %
zero-rw READ 91742 127760 39.26 %
zero-rw WRITE 92501 128818 39.26 %
zero-randrw READ 78925 145437 84.27 %
zero-randrw WRITE 78716 145051 84.27 %
disk-read READ 203705 237220 16.45 %
disk-write WRITE 64068 230051 259.07 %
disk-randread READ 60698 77793 28.16 %
disk-randwrite WRITE 18535 32227 73.87 %
disk-rw READ 49023 65775 34.17 %
disk-rw WRITE 49429 66320 34.17 %
disk-randrw READ 10740 13084 21.82 %
disk-randrw WRITE 10712 13049 21.82 %
Now I would like to work on simulating different workloads (these tests
were simulating 4 jobs in fio on otherwise idling machine)
Ondrej
--
dm-devel mailing list
dm-devel@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/dm-devel
