db-ojb-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jakob Braeuchi <jbraeu...@gmx.ch>
Subject Re: OJB 1.0.x ==> OJB 1.x merge
Date Sun, 28 Aug 2005 08:00:44 GMT
hi armin,

the negative performance impact is imo caused by different versions of 
hsqldb.

jakob

Armin Waibel schrieb:
> ha! it's (almost) done (443 files are handled). I will do the merge of 
> the forrest documentation next week (again, please don't update docs in 
> HEAD till the merge is completed).
> 
> After update CVS HEAD, the OJB-test-suite should give you this result 
> when running against hsql:
> 
> junit-no-compile-no-prepare:
> [junit] Running org.apache.ojb.broker.AllTests
> [junit] Tests run: 726, Failures: 1, Errors: 4, Time elapsed: 194,968 sec
> [junit] Test org.apache.ojb.broker.AllTests FAILED
> [junit] Running org.apache.ojb.odmg.AllTests
> [junit] Tests run: 211, Failures: 0, Errors: 3, Time elapsed: 36,125 sec
> [junit] Test org.apache.ojb.odmg.AllTests FAILED
> [junit] Running org.apache.ojb.soda.AllTests
> [junit] Tests run: 3, Failures: 3, Errors: 0, Time elapsed: 3,297 sec
> [junit] Test org.apache.ojb.soda.AllTests FAILED
> [junit] Running org.apache.ojb.otm.AllTests
> [junit] Tests run: 79, Failures: 0, Errors: 25, Time elapsed: 20,61 sec
> [junit] Test org.apache.ojb.otm.AllTests FAILED
> 
> There are many open tasks to handle. I will post some notes in the next 
> days.
> 
> Now we can start with the 1.1 road map in JIRA.
> 
> *Andrew*
> Could you check the proxy handling in 1.1 (I don't made any changes, but 
> maybe I corrupt a class). To add the TwoLevelCache proxy-all-references 
> feature from 1.0.x you only have to modify method 
> CachingStrategyTwoLevelImpl#materializeFullObject
> and to add a custom property like 'strategy.forceProxies'
> 
> 
> *Martin*
> Could you check the merged ConnectionFactory/connection 
> pooling/fetchSize stuff.
> 
> 
> *Performance profiling*
> Compare with 1.0.x branch the HEAD show a performance impact when using 
> hsql. In 1.0.x the difference between jdbc and PB-api is about 25% 
> (that's very good) in HEAD it's about 200%
> 
> [ojb] ================================================================
> [ojb]         OJB PERFORMANCE TEST SUMMARY
> [ojb] 12 concurrent threads, handle 500 objects per thread
> [ojb]         - performance mode - results per thread
> [ojb] ================================================================
> [ojb]      API  Period   Total   Total  Insert   Fetch  Update  Delete
> [ojb]            [sec]   [sec]     [%]  [msec]  [msec]  [msec]  [msec]
> [ojb] ----------------------------------------------------------------
> [ojb]     JDBC   1.349   1.132     100     826      27     201      78
> [ojb]       PB   3.942   3.548     313    2423      58     752     314
> [ojb]     ODMG   4.929   4.141     366    2396     131     971     642
> [ojb]      OTM   9.359   8.879     784    6600      77    1418     783
> [ojb] ================================================================
> 
> when using DBCP to pool connections with enabled statement pooling the 
> difference will be reduced to 76% (seems hsql 1.7.5 doesn't support 
> statement pooling by default).
> 
> [ojb] ================================================================
> [ojb]         OJB PERFORMANCE TEST SUMMARY
> [ojb] 12 concurrent threads, handle 500 objects per thread
> [ojb]         - performance mode - results per thread
> [ojb] ================================================================
> [ojb]      API  Period   Total   Total  Insert   Fetch  Update  Delete
> [ojb]            [sec]   [sec]     [%]  [msec]  [msec]  [msec]  [msec]
> [ojb] ----------------------------------------------------------------
> [ojb]     JDBC   1.109   0.879     100     561      16     216      84
> [ojb]       PB   2.047   1.549     176     834      56     458     200
> [ojb]     ODMG   3.432   2.407     274    1515      67     482     342
> [ojb]      OTM   4.625   3.714     423    2054     133     856     670
> [ojb] ================================================================
> 
> Anyway this difference is too much, so we should profiling OJB HEAD to 
> find the hot spots.
> 
> 
> regards,
> Armin
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
> For additional commands, e-mail: ojb-dev-help@db.apache.org
> 
> 

---------------------------------------------------------------------
To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
For additional commands, e-mail: ojb-dev-help@db.apache.org


Mime
View raw message