Timeline for Setting Oracle size of row fetches higher makes my app slower?
Current License: CC BY-SA 3.0
27 events
when toggle format | what | by | license | comment | |
---|---|---|---|---|---|
Jan 27, 2015 at 20:53 | comment | added | rogerdpack | See also stackoverflow.com/q/28161049/32453 | |
Jun 5, 2014 at 11:49 | answer | added | tcorteletti | timeline score: 1 | |
Jan 14, 2013 at 11:48 | answer | added | Sumedh | timeline score: 7 | |
Feb 16, 2012 at 21:04 | history | notice removed | daveslab | ||
Feb 16, 2012 at 21:04 | history | bounty ended | daveslab | ||
Feb 16, 2012 at 21:03 | vote | accept | daveslab | ||
Feb 16, 2012 at 19:33 | answer | added | Aprillion | timeline score: 26 | |
Feb 15, 2012 at 20:45 | history | edited | daveslab | CC BY-SA 3.0 |
Re-stated the question
|
Feb 15, 2012 at 20:13 | comment | added | Sap | Have you tried increasing your JVM memmory? HOw many columns are you fetching? | |
Feb 15, 2012 at 14:57 | comment | added | daveslab | @Grrrrr We've done that also. It's definitely the fetch size that affected it because we ran it several times back to back on the same box and saw these clear results. | |
Feb 15, 2012 at 4:53 | comment | added | Sap | Would you like to isolate the problem by running this program once on the same machine as the DB | |
Feb 14, 2012 at 20:45 | comment | added | daveslab | @Grrrrr yes, my next step is to write the test program you are describing. The app server and the DB are on the same network segment. | |
Feb 14, 2012 at 19:32 | answer | added | maximdim | timeline score: 1 | |
Feb 14, 2012 at 19:13 | comment | added | Sap | Your issue might be because of multiple reasons one being network, second being your JVM memory allocation, I am hoping that your DB is not getting accessed by anything else and thus there is no issue on DB. Is your DB on a different machine on a network? in this case TCP might be playing a big role in data transport. Also why don't you write a program which runs your code multiple times with increasing defaultRowPrefetch in each iteration by 10. Compare the performance and choose the optimal prefetchSize for you. | |
Feb 14, 2012 at 19:02 | answer | added | anon | timeline score: 12 | |
Feb 14, 2012 at 15:15 | comment | added | daveslab | Yes, both databases are Oracle, and we're fairly sure that's the issue because we run our job twice (which did the same thing in both cases) and with the default fetch size of 10, it took 10 minutes and with a fetch size of 1000 it took 12 minutes. | |
Feb 14, 2012 at 12:58 | comment | added | tbone | are these 2 databases both Oracle instances? Or Oracle and some other db. If so, how do u know the issue is with pulling data from Oracle? | |
Feb 13, 2012 at 21:36 | comment | added | daveslab | Yeah, because basically we're grabbing data from two different databases and they can't be linked at all. So we need a third party to do the comparison. | |
Feb 13, 2012 at 19:40 | comment | added | Adam Musch | Is there a reason you're not doing that kind of heavy lifting inside the database itself? | |
Feb 13, 2012 at 19:20 | comment | added | daveslab | We're talking reading a few thousand rows several hundred times each, e.g. 10,000 rows * 500 times | |
Feb 13, 2012 at 17:41 | answer | added | Adam Musch | timeline score: 6 | |
Feb 13, 2012 at 17:36 | comment | added | Adam Musch | How long - as in how much data - is each row being fetched, as a rough estimate? | |
Feb 13, 2012 at 17:26 | history | edited | daveslab | CC BY-SA 3.0 |
added 76 characters in body
|
Feb 13, 2012 at 17:13 | history | notice added | daveslab | Draw attention | |
Feb 13, 2012 at 17:13 | history | bounty started | daveslab | ||
Feb 13, 2012 at 17:12 | history | edited | daveslab | CC BY-SA 3.0 |
edited title
|
Feb 9, 2012 at 22:51 | history | asked | daveslab | CC BY-SA 3.0 |