Pg+dSpam: zvyseni vykonu

[Jan Kasprzak]

	Jeste pridam pro srovnani jedno mereni k teto tabulce:

Jan Kasprzak wrote:
: 	Provedl jsem nejaka mereni, jak se ten opozdeny commit
: projevi na vykonu systemu. Jsou to hruba data, ktera jsou jeste navic
: ovlivnena tim, pro ktere uzivatele zrovna chodily maily (prazdna/plna
: databaze dSpamu pro toho uzivatele), a kolik jich chodilo zaroven.
: Urcite to bude zatizeno velkym sumem, nicmene neco z toho videt je:
: 
: Dec 13 14:32:00 delay=100000 siblings=5 421 16.81 29.28 31.93 72.16
: Dec 13 14:38:30 delay=50000 siblings=5 1263 9.54 13.19 21.94 57.90
: Dec 13 14:54:00 delay=25000 siblings=5 1236 5.40 7.57 11.97 29.13
: Dec 13 15:02:30 delay=12500 siblings=5 494 2.86 5.57 7.89 17.81
: Dec 13 15:13:30 delay=6125 siblings=5 502 2.50 4.33 7.33 16.07
: Dec 13 15:24:00 delay=3062 siblings=5 634 2.77 4.12 5.83 12.07
: Dec 13 15:33:00 delay=1531 siblings=5 283 2.02 3.40 4.86 14.54
: Dec 13 15:41:00 delay=766 siblings=5 583 2.45 4.13 6.02 13.33
: Dec 13 15:54:30 delay=383 siblings=5 446 2.76 4.08 6.72 11.48
: Dec 13 16:03:00 delay=192 siblings=5 605 3.72 5.95 9.49 20.85
: Dec 13 16:11:00 delay=96 siblings=5 436 3.62 5.82 10.61 15.98
: Dec 13 16:19:00 delay=48 siblings=5 1539 7.37 8.93 18.18 34.14
: Dec 13 16:29:00 delay=24 siblings=5 597 6.30 11.47 20.87 36.57
: Dec 13 16:37:00 delay=0 siblings=X 612 7.66 16.55 25.32 40.08
: Dec 13 16:47:00 delay=1500 siblings=5 460 2.63 4.29 7.59 16.02
: Dec 13 16:57:00 konec

Dec 13 17:26:00 synchronous_commit=off 502 0.42 0.47 0.56 1.64
Dec 13 17:34:00 konec

Cili tohle je teoreticke maximum propustnosti systemu. Je otazka,
jestli by ho bylo mozne vhodnymi nastavenimi dosahnout.

: 
: Jednotliva pole na radku jsou:
: zacatek mereni (a konec predchoziho mereni), commit_delay, commit_siblings,
: pocet mailu zpracovanych dSpamem v teto periode,
: median doby zpracovani (50 % pripadu ma stejny nebo lepsi cas),
: 3. kvartil doby zpracovani (75 % pripadu ma stejny nebo lepsi cas),
: 9. decil doby zpracovani (90 % --""--),
: 99. percentil doby zpracovani (99 % --""--)

-Yenya

-- 
| Jan "Yenya" Kasprzak  <kas at {fi.muni.cz - work | yenya.net - private}> |
| GPG: ID 1024/D3498839      Fingerprint 0D99A7FB206605D7 8B35FCDE05B18A5E |
| http://www.fi.muni.cz/~kas/    Journal: http://www.fi.muni.cz/~kas/blog/ |
Please don't top post and in particular don't attach entire digests to your
mail or we'll all soon be using bittorrent to read the list.     --Alan Cox