Форум Сайтостроение Хостинг

SAS vs. SSD, Сравнительные тесты дисковых подсистем.

1... 67891011121314 15
[Удален]
#101

ENELIS, FreeBSD 9.1-RELEASE

ENELIS
На сайте с 29.08.2008
Offline
194
ENELIS
#102
WapGraf:
ENELIS, FreeBSD 9.1-RELEASE

Тогда очень странно что write performance низкий настолько.

Вот данные на очень нагруженной системе (полностью, памяти тоже нет)

FreeBSD 8.2 zfs соответственно 15

Файл 256 мегабайт

Кеш райда 256 мегабайт...

конфиг простой zpool поверх hw raid 1

Sequential reads: (g=0): rw=read, bs=4K-4K/4K-4K/4K-4K, ioengine=posixaio, iodepth=16
Random reads: (g=1): rw=randread, bs=4K-4K/4K-4K/4K-4K, ioengine=posixaio, iodepth=16
Sequential writes: (g=2): rw=write, bs=4K-4K/4K-4K/4K-4K, ioengine=posixaio, iodepth=16
Random writes: (g=3): rw=randwrite, bs=4K-4K/4K-4K/4K-4K, ioengine=posixaio, iodepth=16
fio-2.1.1
Starting 4 threads
Sequential reads: Laying out IO file(s) (1 file(s) / 256MB)
Jobs: 1 (f=1): [___w] [62.5% done] [0KB/181.7MB/0KB /s] [0/46.5K/0 iops] [eta 00m:03s]       
Sequential reads: (groupid=0, jobs=1): err= 0: pid=101280: Thu Oct 17 20:54:23 2013
  read : io=262144KB, bw=243628KB/s, iops=60907, runt=  1076msec
    slat (usec): min=1, max=487, avg=11.85, stdev= 8.21
    clat (usec): min=11, max=1343, avg=132.11, stdev=83.53
     lat (usec): min=22, max=1354, avg=143.96, stdev=84.57
    clat percentiles (usec):
     |  1.00th=[   13],  5.00th=[   18], 10.00th=[   31], 20.00th=[   54],
     | 30.00th=[   78], 40.00th=[  103], 50.00th=[  126], 60.00th=[  149],
     | 70.00th=[  173], 80.00th=[  197], 90.00th=[  227], 95.00th=[  266],
     | 99.00th=[  386], 99.50th=[  446], 99.90th=[  620], 99.95th=[  740],
     | 99.99th=[ 1112]
    bw (KB  /s): min=243080, max=247984, per=100.00%, avg=245532.00, stdev=3467.65
    lat (usec) : 20=5.18%, 50=11.39%, 100=22.02%, 250=54.82%, 500=6.29%
    lat (usec) : 750=0.24%, 1000=0.03%
    lat (msec) : 2=0.02%
  cpu          : usr=18.79%, sys=27.81%, ctx=63115, majf=0, minf=0
  IO depths    : 1=5.8%, 2=12.4%, 4=25.1%, 8=50.3%, 16=6.3%, 32=0.0%, >=64=0.0%
     submit    : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
     complete  : 0=0.0%, 4=94.1%, 8=0.0%, 16=5.9%, 32=0.0%, 64=0.0%, >=64=0.0%
     issued    : total=r=65536/w=0/d=0, short=r=0/w=0/d=0
Random reads: (groupid=1, jobs=1): err= 0: pid=101280: Thu Oct 17 20:54:23 2013
  read : io=262144KB, bw=303407KB/s, iops=75851, runt=   864msec
    slat (usec): min=1, max=518, avg= 9.25, stdev= 9.01
    clat (usec): min=13, max=760, avg=114.40, stdev=70.25
     lat (usec): min=23, max=766, avg=123.64, stdev=71.98
    clat percentiles (usec):
     |  1.00th=[   14],  5.00th=[   29], 10.00th=[   39], 20.00th=[   55],
     | 30.00th=[   70], 40.00th=[   86], 50.00th=[  102], 60.00th=[  117],
     | 70.00th=[  137], 80.00th=[  163], 90.00th=[  211], 95.00th=[  247],
     | 99.00th=[  334], 99.50th=[  386], 99.90th=[  506], 99.95th=[  548],
     | 99.99th=[  652]
    bw (KB  /s): min=366632, max=366632, per=100.00%, avg=366632.00, stdev= 0.00
    lat (usec) : 20=2.11%, 50=14.53%, 100=31.94%, 250=46.84%, 500=4.47%
    lat (usec) : 750=0.11%, 1000=0.01%
  cpu          : usr=20.86%, sys=46.81%, ctx=24847, majf=0, minf=0
  IO depths    : 1=2.3%, 2=10.7%, 4=26.7%, 8=53.6%, 16=6.7%, 32=0.0%, >=64=0.0%
     submit    : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
     complete  : 0=0.0%, 4=93.7%, 8=0.1%, 16=6.3%, 32=0.0%, 64=0.0%, >=64=0.0%
     issued    : total=r=65536/w=0/d=0, short=r=0/w=0/d=0
Sequential writes: (groupid=2, jobs=1): err= 0: pid=101280: Thu Oct 17 20:54:23 2013
  write: io=262144KB, bw=237020KB/s, iops=59254, runt=  1106msec
    slat (usec): min=1, max=908, avg=13.16, stdev=15.95
    clat (usec): min=13, max=1480, avg=150.96, stdev=89.10
     lat (usec): min=28, max=1484, avg=164.12, stdev=89.89
    clat percentiles (usec):
     |  1.00th=[   14],  5.00th=[   35], 10.00th=[   53], 20.00th=[   74],
     | 30.00th=[   94], 40.00th=[  114], 50.00th=[  137], 60.00th=[  159],
     | 70.00th=[  185], 80.00th=[  223], 90.00th=[  274], 95.00th=[  310],
     | 99.00th=[  402], 99.50th=[  454], 99.90th=[  612], 99.95th=[  732],
     | 99.99th=[ 1208]
    bw (KB  /s): min=231568, max=246096, per=100.00%, avg=238832.00, stdev=10272.85
    lat (usec) : 20=2.67%, 50=5.47%, 100=24.73%, 250=52.82%, 500=14.00%
    lat (usec) : 750=0.26%, 1000=0.03%
    lat (msec) : 2=0.02%
  cpu          : usr=18.46%, sys=37.01%, ctx=28379, majf=0, minf=0
  IO depths    : 1=2.9%, 2=7.3%, 4=24.9%, 8=57.6%, 16=7.3%, 32=0.0%, >=64=0.0%
     submit    : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
     complete  : 0=0.0%, 4=93.3%, 8=0.2%, 16=6.6%, 32=0.0%, 64=0.0%, >=64=0.0%
     issued    : total=r=0/w=65536/d=0, short=r=0/w=0/d=0
Random writes: (groupid=3, jobs=1): err= 0: pid=101280: Thu Oct 17 20:54:23 2013
  write: io=262144KB, bw=233848KB/s, iops=58462, runt=  1121msec
    slat (usec): min=2, max=935, avg=13.31, stdev=15.23
    clat (usec): min=13, max=1339, avg=150.84, stdev=90.10
     lat (usec): min=27, max=1344, avg=164.15, stdev=91.05
    clat percentiles (usec):
     |  1.00th=[   14],  5.00th=[   34], 10.00th=[   53], 20.00th=[   73],
     | 30.00th=[   93], 40.00th=[  113], 50.00th=[  137], 60.00th=[  159],
     | 70.00th=[  187], 80.00th=[  227], 90.00th=[  282], 95.00th=[  314],
     | 99.00th=[  378], 99.50th=[  422], 99.90th=[  588], 99.95th=[  780],
     | 99.99th=[ 1304]
    bw (KB  /s): min=231504, max=233392, per=99.40%, avg=232448.00, stdev=1335.02
    lat (usec) : 20=2.96%, 50=5.53%, 100=25.32%, 250=50.78%, 500=15.20%
    lat (usec) : 750=0.16%, 1000=0.02%
    lat (msec) : 2=0.04%
  cpu          : usr=9.38%, sys=44.29%, ctx=30635, majf=0, minf=0
  IO depths    : 1=3.1%, 2=7.3%, 4=24.9%, 8=57.5%, 16=7.2%, 32=0.0%, >=64=0.0%
     submit    : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
     complete  : 0=0.0%, 4=93.3%, 8=0.2%, 16=6.6%, 32=0.0%, 64=0.0%, >=64=0.0%
     issued    : total=r=0/w=65536/d=0, short=r=0/w=0/d=0

Run status group 0 (all jobs):
   READ: io=262144KB, aggrb=243628KB/s, minb=243628KB/s, maxb=243628KB/s, mint=1076msec, maxt=1076msec

Run status group 1 (all jobs):
   READ: io=262144KB, aggrb=303407KB/s, minb=303407KB/s, maxb=303407KB/s, mint=864msec, maxt=864msec

Run status group 2 (all jobs):
  WRITE: io=262144KB, aggrb=237019KB/s, minb=237019KB/s, maxb=237019KB/s, mint=1106msec, maxt=1106msec

Run status group 3 (all jobs):
  WRITE: io=262144KB, aggrb=233848KB/s, minb=233848KB/s, maxb=233848KB/s, mint=1121msec, maxt=1121msec
С Уважением, ServerAstra.ru (https://serverastra.com) - VPS и выделенные сервера в Будапеште по выгодным ценам!
FreeBSD, монтирование рейда на Сервер с SAS|SSD дисками хостинги гарантирующие сохранность данных
[Удален]
#103

ENELIS, меня также это удивило. Но сравнивать буду линуксовые системы и с равным количеством дисков. Для большей адекватности тестов.

ENELIS
На сайте с 29.08.2008
Offline
194
ENELIS
#104

Вот на ней же 1536 мегабайт, чтобы в кеши не влезало, в принципе мы видим реальную ситуацию с random writes.

Если что это RAID 10 из 4 дисков по 146гб 10k rpm.

Sequential reads: (g=0): rw=read, bs=4K-4K/4K-4K/4K-4K, ioengine=posixaio, iodepth=16
Random reads: (g=1): rw=randread, bs=4K-4K/4K-4K/4K-4K, ioengine=posixaio, iodepth=16
Sequential writes: (g=2): rw=write, bs=4K-4K/4K-4K/4K-4K, ioengine=posixaio, iodepth=16
Random writes: (g=3): rw=randwrite, bs=4K-4K/4K-4K/4K-4K, ioengine=posixaio, iodepth=16
fio-2.1.1
Starting 4 threads
Sequential reads: Laying out IO file(s) (1 file(s) / 1536MB)
Jobs: 1 (f=1): [___w] [97.3% done] [0KB/248KB/0KB /s] [0/62/0 iops] [eta 00m:23s]      
Sequential reads: (groupid=0, jobs=1): err= 0: pid=101280: Thu Oct 17 21:16:13 2013
  read : io=1536.0MB, bw=205603KB/s, iops=51400, runt=  7650msec
    slat (usec): min=1, max=157162, avg=15.03, stdev=440.70
    clat (usec): min=11, max=458218, avg=158.15, stdev=1662.10
     lat (usec): min=21, max=458221, avg=173.18, stdev=1721.49
    clat percentiles (usec):
     |  1.00th=[   13],  5.00th=[   26], 10.00th=[   35], 20.00th=[   54],
     | 30.00th=[   72], 40.00th=[   92], 50.00th=[  110], 60.00th=[  131],
     | 70.00th=[  155], 80.00th=[  183], 90.00th=[  221], 95.00th=[  278],
     | 99.00th=[  564], 99.50th=[  964], 99.90th=[ 4048], 99.95th=[11456],
     | 99.99th=[85504]
    bw (KB  /s): min=27033, max=354312, per=99.56%, avg=204698.73, stdev=96811.42
    lat (usec) : 20=4.07%, 50=13.24%, 100=26.85%, 250=49.22%, 500=5.40%
    lat (usec) : 750=0.55%, 1000=0.19%
    lat (msec) : 2=0.28%, 4=0.10%, 10=0.04%, 20=0.03%, 50=0.02%
    lat (msec) : 100=0.01%, 250=0.01%, 500=0.01%
  cpu          : usr=7.77%, sys=33.14%, ctx=273751, majf=0, minf=0
  IO depths    : 1=3.9%, 2=11.2%, 4=26.0%, 8=52.3%, 16=6.6%, 32=0.0%, >=64=0.0%
     submit    : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
     complete  : 0=0.0%, 4=93.9%, 8=0.1%, 16=6.1%, 32=0.0%, 64=0.0%, >=64=0.0%
     issued    : total=r=393216/w=0/d=0, short=r=0/w=0/d=0
Random reads: (groupid=1, jobs=1): err= 0: pid=101280: Thu Oct 17 21:16:13 2013
  read : io=1536.0MB, bw=181666KB/s, iops=45416, runt=  8658msec
    slat (usec): min=2, max=1501.1K, avg=17.77, stdev=2403.34
    clat (usec): min=11, max=1502.3K, avg=163.63, stdev=4196.58
     lat (usec): min=23, max=1502.7K, avg=181.40, stdev=4836.62
    clat percentiles (usec):
     |  1.00th=[   14],  5.00th=[   30], 10.00th=[   41], 20.00th=[   60],
     | 30.00th=[   79], 40.00th=[   99], 50.00th=[  119], 60.00th=[  145],
     | 70.00th=[  173], 80.00th=[  217], 90.00th=[  270], 95.00th=[  334],
     | 99.00th=[  540], 99.50th=[  668], 99.90th=[ 1592], 99.95th=[ 2768],
     | 99.99th=[23680]
    bw (KB  /s): min= 8765, max=341208, per=100.00%, avg=208128.40, stdev=98893.63
    lat (usec) : 20=3.24%, 50=11.69%, 100=25.21%, 250=46.40%, 500=12.22%
    lat (usec) : 750=0.90%, 1000=0.16%
    lat (msec) : 2=0.11%, 4=0.05%, 10=0.02%, 50=0.01%, 100=0.01%
    lat (msec) : 250=0.01%, 2000=0.01%
  cpu          : usr=10.33%, sys=29.38%, ctx=226297, majf=0, minf=0
  IO depths    : 1=3.5%, 2=10.4%, 4=26.3%, 8=53.1%, 16=6.6%, 32=0.0%, >=64=0.0%
     submit    : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
     complete  : 0=0.0%, 4=93.8%, 8=0.1%, 16=6.2%, 32=0.0%, 64=0.0%, >=64=0.0%
     issued    : total=r=393216/w=0/d=0, short=r=0/w=0/d=0
Sequential writes: (groupid=2, jobs=1): err= 0: pid=101280: Thu Oct 17 21:16:13 2013
  write: io=1536.0MB, bw=44670KB/s, iops=11167, runt= 35211msec
    slat (usec): min=1, max=644645, avg=36.35, stdev=1422.03
    clat (usec): min=9, max=6385.9K, avg=1127.88, stdev=44379.24
     lat (usec): min=25, max=6385.9K, avg=1164.23, stdev=44409.66
    clat percentiles (usec):
     |  1.00th=[   13],  5.00th=[   15], 10.00th=[   34], 20.00th=[   68],
     | 30.00th=[  102], 40.00th=[  129], 50.00th=[  161], 60.00th=[  193],
     | 70.00th=[  231], 80.00th=[  266], 90.00th=[  334], 95.00th=[  908],
     | 99.00th=[ 5536], 99.50th=[11200], 99.90th=[156672], 99.95th=[250880],
     | 99.99th=[1036288]
    bw (KB  /s): min=   36, max=216896, per=100.00%, avg=60334.73, stdev=58099.83
    lat (usec) : 10=0.01%, 20=5.50%, 50=7.39%, 100=16.51%, 250=46.05%
    lat (usec) : 500=17.50%, 750=1.44%, 1000=1.07%
    lat (msec) : 2=1.78%, 4=1.36%, 10=0.86%, 20=0.19%, 50=0.13%
    lat (msec) : 100=0.06%, 250=0.10%, 500=0.03%, 750=0.01%, 1000=0.01%
    lat (msec) : 2000=0.01%, >=2000=0.01%
  cpu          : usr=2.78%, sys=5.98%, ctx=346971, majf=0, minf=0
  IO depths    : 1=4.5%, 2=10.7%, 4=23.6%, 8=53.7%, 16=7.5%, 32=0.0%, >=64=0.0%
     submit    : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
     complete  : 0=0.0%, 4=93.8%, 8=0.2%, 16=6.0%, 32=0.0%, 64=0.0%, >=64=0.0%
     issued    : total=r=0/w=393216/d=0, short=r=0/w=0/d=0
Random writes: (groupid=3, jobs=1): err= 0: pid=101214: Thu Oct 17 21:16:13 2013
  write: io=1536.0MB, bw=2062.6KB/s, iops=515, runt=762601msec
    slat (usec): min=1, max=580663, avg=159.66, stdev=2470.08
    clat (usec): min=6, max=4596.7K, avg=29807.89, stdev=218886.77
     lat (usec): min=23, max=4596.7K, avg=29967.56, stdev=218911.52
    clat percentiles (usec):
     |  1.00th=[   14],  5.00th=[   27], 10.00th=[   70], 20.00th=[  215],
     | 30.00th=[  350], 40.00th=[  490], 50.00th=[  636], 60.00th=[  812],
     | 70.00th=[ 1064], 80.00th=[ 1848], 90.00th=[20096], 95.00th=[104960],
     | 99.00th=[460800], 99.50th=[1433600], 99.90th=[3522560], 99.95th=[4046848],
     | 99.99th=[4423680]
    bw (KB  /s): min=    2, max=28717, per=100.00%, avg=3474.92, stdev=6039.38
    lat (usec) : 10=0.07%, 20=4.59%, 50=3.31%, 100=4.23%, 250=10.49%
    lat (usec) : 500=18.04%, 750=16.06%, 1000=11.35%
    lat (msec) : 2=12.62%, 4=3.98%, 10=3.16%, 20=2.09%, 50=2.62%
    lat (msec) : 100=2.20%, 250=3.16%, 500=1.09%, 750=0.24%, 1000=0.09%
    lat (msec) : 2000=0.19%, >=2000=0.41%
  cpu          : usr=0.19%, sys=0.31%, ctx=361515, majf=0, minf=0
  IO depths    : 1=4.6%, 2=10.2%, 4=22.6%, 8=53.7%, 16=8.9%, 32=0.0%, >=64=0.0%
     submit    : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
     complete  : 0=0.0%, 4=93.9%, 8=0.6%, 16=5.5%, 32=0.0%, 64=0.0%, >=64=0.0%
     issued    : total=r=0/w=393216/d=0, short=r=0/w=0/d=0

Run status group 0 (all jobs):
   READ: io=1536.0MB, aggrb=205603KB/s, minb=205603KB/s, maxb=205603KB/s, mint=7650msec, maxt=7650msec

Run status group 1 (all jobs):
   READ: io=1536.0MB, aggrb=181665KB/s, minb=181665KB/s, maxb=181665KB/s, mint=8658msec, maxt=8658msec

Run status group 2 (all jobs):
  WRITE: io=1536.0MB, aggrb=44669KB/s, minb=44669KB/s, maxb=44669KB/s, mint=35211msec, maxt=35211msec

Run status group 3 (all jobs):
  WRITE: io=1536.0MB, aggrb=2062KB/s, minb=2062KB/s, maxb=2062KB/s, mint=762601msec, maxt=762601msec
Издатели AdSense жалуются на Факторы ранжирования Google в Google выделяет 10 Мегабайт
[Удален]
#105

ENELIS, вот смотрите, на том же сервере (с теми же дисками) добился вот такой картинки:

Sequential reads: bw=313945KB/s, iops=78486
Random reads: bw=355209KB/s, iops=88802
Sequential writes: bw=337814KB/s, iops=84453
Random writes: bw=331408KB/s, iops=82852

Вот по этому адекватнее использовать в тестах линукс)

LineHost
На сайте с 20.01.2007
Offline
339
LineHost
#106

Вы что то запутались с этими тестами, я ни за белых ни за красных, но будущее вижу только с SSD + SATA. SAS гиблое, мёртвое и уже никому ненужное дорогое прошлое. Я хоть и сторонник ретро стиля, но не до такого уровня ;)

Неужели думаете что кто нибудь осилит внимательно изучить все ваши логи? И ещё думать где ошибки и почему такие разные результаты? Кому это надо? Берёте техническую документацию производителя и налицо все параметры.... Я перекидываю две впски, которые в сумме утилизируает в среднем на 60% SAS на SSD и получаю утилизацию SSD ~0.3-0.5%. И это уже из трёхлетнего опыта использования SSD... Именя возможность в случае надобности использовать SSD можно забыть о проблемах с IO, по крайней мере на сегодняшний день....

В общем, клиентам вобще по барабану на каких дисках крутится ихние ресурсы, им главное как крутится ;)

SERV.LT - Стабильные услуги хостинга, KVM VPS в Литве, Франции. (https://www.serv.lt/ru/vps/kvm/) Недорогие выделенные серверы (https://www.serv.lt/ru/dedicated-lt/) в Литве.
тест производительности. VDS/VPS vs Разница ВДС хостинга на VPS (Развод или нет?)
[Удален]
#107

LineHost, уж быстрее SSD вырастут до приличных размеров нежели за SATA станет будущее!

P
На сайте с 16.03.2009
Offline
144
poiuty
#108
WapGraf:
LineHost, уж быстрее SSD вырастут до приличных размеров нежели за SATA станет будущее!

Hetzner, уже 480гб ссд за 39 € monthly (vat)

http://wiki.hetzner.de/index.php/Preisliste_Zusatzprodukte/en

http://www.ramcity.com.au/buy/samsung-sm843t-480gb-2.5-inch-7mm-ssd/MZ7WD480HAGM-00003#tab_specifications

L
На сайте с 12.06.2013
Offline
3
larvaq
#109

clat percentiles (usec):

| 1.00th=[ 14], 5.00th=[ 34], 10.00th=[ 53], 20.00th=[ 73],

| 30.00th=[ 93], 40.00th=[ 113], 50.00th=[ 137], 60.00th=[ 159],

| 70.00th=[ 187], 80.00th=[ 227], 90.00th=[ 282], 95.00th=[ 314],

| 99.00th=[ 378], 99.50th=[ 422], 99.90th=[ 588], 99.95th=[ 780],

ошибка тут

Что ищут люди сегодня: Методика Н. Харина-И.Ашманова для Грамотное сегментирование объявлений -
[Удален]
#110

poiuty, полугиговые давно на рынке уже.

Полагаю LineHost'у нужно more more ....

Вот как слепят SSD на 2-3ТБ так сата мхом порастут.

1... 67891011121314 15

Авторизуйтесь или зарегистрируйтесь, чтобы оставить комментарий