You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@jena.apache.org by "Simon Helsen (JIRA)" <ji...@apache.org> on 2012/06/14 17:48:42 UTC
[jira] [Created] (JENA-256) Significant performance regression
(TDB?) on 2.7.1 RC compared to May 15 build
Simon Helsen created JENA-256:
---------------------------------
Summary: Significant performance regression (TDB?) on 2.7.1 RC compared to May 15 build
Key: JENA-256
URL: https://issues.apache.org/jira/browse/JENA-256
Project: Apache Jena
Issue Type: Bug
Components: TDB
Affects Versions: Jena 2.7.1
Environment: Windows 7, 64 bit, tested against 2.7.1 RC from June 9 versus a build from May 15
Reporter: Simon Helsen
Priority: Critical
See also http://mail-archives.apache.org/mod_mbox/jena-dev/201206.mbox/%3C4FD9F19E.3080904%40bristol.ac.uk%3E
I was able to reproduce the performance regression with an isolated test scenario. So I recreated the components ARQ, CORE, IRI, and TDB with the SVN state of May 15 9 (so svn update -r {2012-05-15} and then svn clean install)
I then created a simple test program (attached as PerformanceRegressionTest.java) which I ran 4 times in a row for each version of Jena. Note that I deleted the TDB directory after the first 4 runs before using the other Jena version. Attached are the files with the output. The regression is obvious
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Updated] (JENA-256) Significant performance regression
(TDB?) on 2.7.1 RC compared to May 15 build
Posted by "Simon Helsen (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/JENA-256?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Simon Helsen updated JENA-256:
------------------------------
Attachment: 271RCOutput4.txt
271RCOutput3.txt
271RCOutput2.txt
271RCOutput1.txt
271May15Output4.txt
271May15Output3.txt
271May15Output2.txt
271May15Output1.txt
PerformanceRegressionTest.java
> Significant performance regression (TDB?) on 2.7.1 RC compared to May 15 build
> ------------------------------------------------------------------------------
>
> Key: JENA-256
> URL: https://issues.apache.org/jira/browse/JENA-256
> Project: Apache Jena
> Issue Type: Bug
> Components: TDB
> Affects Versions: Jena 2.7.1
> Environment: Windows 7, 64 bit, tested against 2.7.1 RC from June 9 versus a build from May 15
> Reporter: Simon Helsen
> Priority: Critical
> Attachments: 271May15Output1.txt, 271May15Output2.txt, 271May15Output3.txt, 271May15Output4.txt, 271RCOutput1.txt, 271RCOutput2.txt, 271RCOutput3.txt, 271RCOutput4.txt, PerformanceRegressionTest.java
>
>
> See also http://mail-archives.apache.org/mod_mbox/jena-dev/201206.mbox/%3C4FD9F19E.3080904%40bristol.ac.uk%3E
> I was able to reproduce the performance regression with an isolated test scenario. So I recreated the components ARQ, CORE, IRI, and TDB with the SVN state of May 15 9 (so svn update -r {2012-05-15} and then svn clean install)
> I then created a simple test program (attached as PerformanceRegressionTest.java) which I ran 4 times in a row for each version of Jena. Note that I deleted the TDB directory after the first 4 runs before using the other Jena version. Attached are the files with the output. The regression is obvious
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Updated] (JENA-256) Significant performance regression
(TDB?) on 2.7.1 RC compared to May 15 build
Posted by "Simon Helsen (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/JENA-256?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Simon Helsen updated JENA-256:
------------------------------
Attachment: 271May15Output4.txt
271May15Output3.txt
271May15Output2.txt
made a copy mistake
> Significant performance regression (TDB?) on 2.7.1 RC compared to May 15 build
> ------------------------------------------------------------------------------
>
> Key: JENA-256
> URL: https://issues.apache.org/jira/browse/JENA-256
> Project: Apache Jena
> Issue Type: Bug
> Components: TDB
> Affects Versions: Jena 2.7.1
> Environment: Windows 7, 64 bit, tested against 2.7.1 RC from June 9 versus a build from May 15
> Reporter: Simon Helsen
> Priority: Critical
> Attachments: 271May15Output1.txt, 271May15Output2.txt, 271May15Output3.txt, 271May15Output4.txt, 271RCOutput1.txt, 271RCOutput2.txt, 271RCOutput3.txt, 271RCOutput4.txt, PerformanceRegressionTest.java
>
>
> See also http://mail-archives.apache.org/mod_mbox/jena-dev/201206.mbox/%3C4FD9F19E.3080904%40bristol.ac.uk%3E
> I was able to reproduce the performance regression with an isolated test scenario. So I recreated the components ARQ, CORE, IRI, and TDB with the SVN state of May 15 9 (so svn update -r {2012-05-15} and then svn clean install)
> I then created a simple test program (attached as PerformanceRegressionTest.java) which I ran 4 times in a row for each version of Jena. Note that I deleted the TDB directory after the first 4 runs before using the other Jena version. Attached are the files with the output. The regression is obvious
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (JENA-256) Significant performance regression
(TDB?) on 2.7.1 RC compared to May 15 build
Posted by "Simon Helsen (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/JENA-256?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13418331#comment-13418331 ]
Simon Helsen commented on JENA-256:
-----------------------------------
if the codebase has it on, I'll just regenerate a local snapshot
> Significant performance regression (TDB?) on 2.7.1 RC compared to May 15 build
> ------------------------------------------------------------------------------
>
> Key: JENA-256
> URL: https://issues.apache.org/jira/browse/JENA-256
> Project: Apache Jena
> Issue Type: Bug
> Components: TDB
> Affects Versions: Jena 2.7.1
> Environment: Windows 7, 64 bit, tested against 2.7.1 RC from June 9 versus a build from May 15
> Reporter: Simon Helsen
> Assignee: Andy Seaborne
> Priority: Critical
> Fix For: TDB 0.9.3
>
> Attachments: 271May15Output1.txt, 271May15Output2.txt, 271May15Output3.txt, 271May15Output4.txt, 271RCOutput1.txt, 271RCOutput2.txt, 271RCOutput3.txt, 271RCOutput4.txt, PerformanceRegressionTest.java
>
>
> See also http://mail-archives.apache.org/mod_mbox/jena-dev/201206.mbox/%3C4FD9F19E.3080904%40bristol.ac.uk%3E
> I was able to reproduce the performance regression with an isolated test scenario. So I recreated the components ARQ, CORE, IRI, and TDB with the SVN state of May 15 9 (so svn update -r {2012-05-15} and then svn clean install)
> I then created a simple test program (attached as PerformanceRegressionTest.java) which I ran 4 times in a row for each version of Jena. Note that I deleted the TDB directory after the first 4 runs before using the other Jena version. Attached are the files with the output. The regression is obvious
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (JENA-256) Significant performance regression
(TDB?) on 2.7.1 RC compared to May 15 build
Posted by "Simon Helsen (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/JENA-256?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13295174#comment-13295174 ]
Simon Helsen commented on JENA-256:
-----------------------------------
so I tested it with Sun's JRE and the effect is more or less the same. It is slighly less outspoken, but still clearly slower. One thing to note that is the first test run with RC is not as slow as subsequent ones, which seems different with the May 15 build. However, even the first run with the RC candidate is always slower than the first run with the May 15 build, whether it runs on IBM's or Oracle's JRE
> Significant performance regression (TDB?) on 2.7.1 RC compared to May 15 build
> ------------------------------------------------------------------------------
>
> Key: JENA-256
> URL: https://issues.apache.org/jira/browse/JENA-256
> Project: Apache Jena
> Issue Type: Bug
> Components: TDB
> Affects Versions: Jena 2.7.1
> Environment: Windows 7, 64 bit, tested against 2.7.1 RC from June 9 versus a build from May 15
> Reporter: Simon Helsen
> Priority: Critical
> Attachments: 271May15Output1.txt, 271May15Output2.txt, 271May15Output3.txt, 271May15Output4.txt, 271RCOutput1.txt, 271RCOutput2.txt, 271RCOutput3.txt, 271RCOutput4.txt, PerformanceRegressionTest.java
>
>
> See also http://mail-archives.apache.org/mod_mbox/jena-dev/201206.mbox/%3C4FD9F19E.3080904%40bristol.ac.uk%3E
> I was able to reproduce the performance regression with an isolated test scenario. So I recreated the components ARQ, CORE, IRI, and TDB with the SVN state of May 15 9 (so svn update -r {2012-05-15} and then svn clean install)
> I then created a simple test program (attached as PerformanceRegressionTest.java) which I ran 4 times in a row for each version of Jena. Note that I deleted the TDB directory after the first 4 runs before using the other Jena version. Attached are the files with the output. The regression is obvious
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (JENA-256) Significant performance regression
(TDB?) on 2.7.1 RC compared to May 15 build
Posted by "Andy Seaborne (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/JENA-256?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13393587#comment-13393587 ]
Andy Seaborne commented on JENA-256:
------------------------------------
Simon - thnaks for the test case and the details.
> Significant performance regression (TDB?) on 2.7.1 RC compared to May 15 build
> ------------------------------------------------------------------------------
>
> Key: JENA-256
> URL: https://issues.apache.org/jira/browse/JENA-256
> Project: Apache Jena
> Issue Type: Bug
> Components: TDB
> Affects Versions: Jena 2.7.1
> Environment: Windows 7, 64 bit, tested against 2.7.1 RC from June 9 versus a build from May 15
> Reporter: Simon Helsen
> Priority: Critical
> Attachments: 271May15Output1.txt, 271May15Output2.txt, 271May15Output3.txt, 271May15Output4.txt, 271RCOutput1.txt, 271RCOutput2.txt, 271RCOutput3.txt, 271RCOutput4.txt, PerformanceRegressionTest.java
>
>
> See also http://mail-archives.apache.org/mod_mbox/jena-dev/201206.mbox/%3C4FD9F19E.3080904%40bristol.ac.uk%3E
> I was able to reproduce the performance regression with an isolated test scenario. So I recreated the components ARQ, CORE, IRI, and TDB with the SVN state of May 15 9 (so svn update -r {2012-05-15} and then svn clean install)
> I then created a simple test program (attached as PerformanceRegressionTest.java) which I ran 4 times in a row for each version of Jena. Note that I deleted the TDB directory after the first 4 runs before using the other Jena version. Attached are the files with the output. The regression is obvious
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (JENA-256) Significant performance regression
(TDB?) on 2.7.1 RC compared to May 15 build
Posted by "Andy Seaborne (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/JENA-256?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13418326#comment-13418326 ]
Andy Seaborne commented on JENA-256:
------------------------------------
The codebase only has the optimization on. It requires a code change to set it to zero batch size.
The batch size is set in TransactionManager.TSM_WriteBackEndTxn.QueueBatchSize to 10.
More than 10 didn't make a great deal of difference.
Ideally, flush-back would all handled on another thread - batching is a cautious step in thet direction.
I used;
apache-jena-2.7.3-20120718.212752-16.zip
and the development codebase.
It is conceivable that it was not in the first 7/18 build because of Jenkins overload.
If that does not make a difference, it is worth trying your example code with a singe read transaction around the whole loop. This causes a similar effect even in the old code.
> Significant performance regression (TDB?) on 2.7.1 RC compared to May 15 build
> ------------------------------------------------------------------------------
>
> Key: JENA-256
> URL: https://issues.apache.org/jira/browse/JENA-256
> Project: Apache Jena
> Issue Type: Bug
> Components: TDB
> Affects Versions: Jena 2.7.1
> Environment: Windows 7, 64 bit, tested against 2.7.1 RC from June 9 versus a build from May 15
> Reporter: Simon Helsen
> Assignee: Andy Seaborne
> Priority: Critical
> Fix For: TDB 0.9.3
>
> Attachments: 271May15Output1.txt, 271May15Output2.txt, 271May15Output3.txt, 271May15Output4.txt, 271RCOutput1.txt, 271RCOutput2.txt, 271RCOutput3.txt, 271RCOutput4.txt, PerformanceRegressionTest.java
>
>
> See also http://mail-archives.apache.org/mod_mbox/jena-dev/201206.mbox/%3C4FD9F19E.3080904%40bristol.ac.uk%3E
> I was able to reproduce the performance regression with an isolated test scenario. So I recreated the components ARQ, CORE, IRI, and TDB with the SVN state of May 15 9 (so svn update -r {2012-05-15} and then svn clean install)
> I then created a simple test program (attached as PerformanceRegressionTest.java) which I ran 4 times in a row for each version of Jena. Note that I deleted the TDB directory after the first 4 runs before using the other Jena version. Attached are the files with the output. The regression is obvious
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Comment Edited] (JENA-256) Significant performance
regression (TDB?) on 2.7.1 RC compared to May 15 build
Posted by "Andy Seaborne (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/JENA-256?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13393587#comment-13393587 ]
Andy Seaborne edited comment on JENA-256 at 6/17/12 7:16 PM:
-------------------------------------------------------------
Simon - thanks for the test case and the details.
was (Author: andy.seaborne):
Simon - thnaks for the test case and the details.
> Significant performance regression (TDB?) on 2.7.1 RC compared to May 15 build
> ------------------------------------------------------------------------------
>
> Key: JENA-256
> URL: https://issues.apache.org/jira/browse/JENA-256
> Project: Apache Jena
> Issue Type: Bug
> Components: TDB
> Affects Versions: Jena 2.7.1
> Environment: Windows 7, 64 bit, tested against 2.7.1 RC from June 9 versus a build from May 15
> Reporter: Simon Helsen
> Priority: Critical
> Attachments: 271May15Output1.txt, 271May15Output2.txt, 271May15Output3.txt, 271May15Output4.txt, 271RCOutput1.txt, 271RCOutput2.txt, 271RCOutput3.txt, 271RCOutput4.txt, PerformanceRegressionTest.java
>
>
> See also http://mail-archives.apache.org/mod_mbox/jena-dev/201206.mbox/%3C4FD9F19E.3080904%40bristol.ac.uk%3E
> I was able to reproduce the performance regression with an isolated test scenario. So I recreated the components ARQ, CORE, IRI, and TDB with the SVN state of May 15 9 (so svn update -r {2012-05-15} and then svn clean install)
> I then created a simple test program (attached as PerformanceRegressionTest.java) which I ran 4 times in a row for each version of Jena. Note that I deleted the TDB directory after the first 4 runs before using the other Jena version. Attached are the files with the output. The regression is obvious
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Updated] (JENA-256) Significant performance regression
(TDB?) on 2.7.1 RC compared to May 15 build
Posted by "Simon Helsen (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/JENA-256?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Simon Helsen updated JENA-256:
------------------------------
Attachment: (was: PerformanceRegressionTest.java)
> Significant performance regression (TDB?) on 2.7.1 RC compared to May 15 build
> ------------------------------------------------------------------------------
>
> Key: JENA-256
> URL: https://issues.apache.org/jira/browse/JENA-256
> Project: Apache Jena
> Issue Type: Bug
> Components: TDB
> Affects Versions: Jena 2.7.1
> Environment: Windows 7, 64 bit, tested against 2.7.1 RC from June 9 versus a build from May 15
> Reporter: Simon Helsen
> Priority: Critical
> Attachments: 271May15Output1.txt, 271May15Output2.txt, 271May15Output3.txt, 271May15Output4.txt, 271RCOutput1.txt, 271RCOutput2.txt, 271RCOutput3.txt, 271RCOutput4.txt, PerformanceRegressionTest.java
>
>
> See also http://mail-archives.apache.org/mod_mbox/jena-dev/201206.mbox/%3C4FD9F19E.3080904%40bristol.ac.uk%3E
> I was able to reproduce the performance regression with an isolated test scenario. So I recreated the components ARQ, CORE, IRI, and TDB with the SVN state of May 15 9 (so svn update -r {2012-05-15} and then svn clean install)
> I then created a simple test program (attached as PerformanceRegressionTest.java) which I ran 4 times in a row for each version of Jena. Note that I deleted the TDB directory after the first 4 runs before using the other Jena version. Attached are the files with the output. The regression is obvious
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (JENA-256) Significant performance regression
(TDB?) on 2.7.1 RC compared to May 15 build
Posted by "Simon Helsen (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/JENA-256?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13295177#comment-13295177 ]
Simon Helsen commented on JENA-256:
-----------------------------------
ok, so I also ran the tests with the createDataset outside the loop. The effect is identical. Although this situation increases the overall speed slightly (in both builds), it hardly makes a difference. This doesn't surprise me because in my first version of the test, I had recorded the time of the dataset creation and (as hoped) it was very small, almost negligible. Another reason I didn't think this would matter was that my previous internal tests showed that non-transactional code was still fast and I need to create a dataset in the non-transactional case as well. In other words, the createDataset is not part of the problem. The regression sits direct in the transaction
> Significant performance regression (TDB?) on 2.7.1 RC compared to May 15 build
> ------------------------------------------------------------------------------
>
> Key: JENA-256
> URL: https://issues.apache.org/jira/browse/JENA-256
> Project: Apache Jena
> Issue Type: Bug
> Components: TDB
> Affects Versions: Jena 2.7.1
> Environment: Windows 7, 64 bit, tested against 2.7.1 RC from June 9 versus a build from May 15
> Reporter: Simon Helsen
> Priority: Critical
> Attachments: 271May15Output1.txt, 271May15Output2.txt, 271May15Output3.txt, 271May15Output4.txt, 271RCOutput1.txt, 271RCOutput2.txt, 271RCOutput3.txt, 271RCOutput4.txt, PerformanceRegressionTest.java
>
>
> See also http://mail-archives.apache.org/mod_mbox/jena-dev/201206.mbox/%3C4FD9F19E.3080904%40bristol.ac.uk%3E
> I was able to reproduce the performance regression with an isolated test scenario. So I recreated the components ARQ, CORE, IRI, and TDB with the SVN state of May 15 9 (so svn update -r {2012-05-15} and then svn clean install)
> I then created a simple test program (attached as PerformanceRegressionTest.java) which I ran 4 times in a row for each version of Jena. Note that I deleted the TDB directory after the first 4 runs before using the other Jena version. Attached are the files with the output. The regression is obvious
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (JENA-256) Significant performance regression
(TDB?) on 2.7.1 RC compared to May 15 build
Posted by "Simon Helsen (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/JENA-256?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13415736#comment-13415736 ]
Simon Helsen commented on JENA-256:
-----------------------------------
thanks for the fix Andy. I will try it out and see how it affects the performance. Provided we find this significantly improves the write performance, do you think it would be reasonable to request a Jena 2.7.3 release?
> Significant performance regression (TDB?) on 2.7.1 RC compared to May 15 build
> ------------------------------------------------------------------------------
>
> Key: JENA-256
> URL: https://issues.apache.org/jira/browse/JENA-256
> Project: Apache Jena
> Issue Type: Bug
> Components: TDB
> Affects Versions: Jena 2.7.1
> Environment: Windows 7, 64 bit, tested against 2.7.1 RC from June 9 versus a build from May 15
> Reporter: Simon Helsen
> Assignee: Andy Seaborne
> Priority: Critical
> Fix For: TDB 0.9.3
>
> Attachments: 271May15Output1.txt, 271May15Output2.txt, 271May15Output3.txt, 271May15Output4.txt, 271RCOutput1.txt, 271RCOutput2.txt, 271RCOutput3.txt, 271RCOutput4.txt, PerformanceRegressionTest.java
>
>
> See also http://mail-archives.apache.org/mod_mbox/jena-dev/201206.mbox/%3C4FD9F19E.3080904%40bristol.ac.uk%3E
> I was able to reproduce the performance regression with an isolated test scenario. So I recreated the components ARQ, CORE, IRI, and TDB with the SVN state of May 15 9 (so svn update -r {2012-05-15} and then svn clean install)
> I then created a simple test program (attached as PerformanceRegressionTest.java) which I ran 4 times in a row for each version of Jena. Note that I deleted the TDB directory after the first 4 runs before using the other Jena version. Attached are the files with the output. The regression is obvious
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Resolved] (JENA-256) Significant performance regression
(TDB?) on 2.7.1 RC compared to May 15 build
Posted by "Andy Seaborne (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/JENA-256?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Andy Seaborne resolved JENA-256.
--------------------------------
Resolution: Fixed
Fix Version/s: TDB 0.9.3
Assignee: Andy Seaborne
Improvements to the TDB write/commit (no readers) path improve the performance.
You will see non-empty journal files more frequently when the system stops even for a writer-only setup. This is now correct behaviour. The journal is flushed on next start-up.
> Significant performance regression (TDB?) on 2.7.1 RC compared to May 15 build
> ------------------------------------------------------------------------------
>
> Key: JENA-256
> URL: https://issues.apache.org/jira/browse/JENA-256
> Project: Apache Jena
> Issue Type: Bug
> Components: TDB
> Affects Versions: Jena 2.7.1
> Environment: Windows 7, 64 bit, tested against 2.7.1 RC from June 9 versus a build from May 15
> Reporter: Simon Helsen
> Assignee: Andy Seaborne
> Priority: Critical
> Fix For: TDB 0.9.3
>
> Attachments: 271May15Output1.txt, 271May15Output2.txt, 271May15Output3.txt, 271May15Output4.txt, 271RCOutput1.txt, 271RCOutput2.txt, 271RCOutput3.txt, 271RCOutput4.txt, PerformanceRegressionTest.java
>
>
> See also http://mail-archives.apache.org/mod_mbox/jena-dev/201206.mbox/%3C4FD9F19E.3080904%40bristol.ac.uk%3E
> I was able to reproduce the performance regression with an isolated test scenario. So I recreated the components ARQ, CORE, IRI, and TDB with the SVN state of May 15 9 (so svn update -r {2012-05-15} and then svn clean install)
> I then created a simple test program (attached as PerformanceRegressionTest.java) which I ran 4 times in a row for each version of Jena. Note that I deleted the TDB directory after the first 4 runs before using the other Jena version. Attached are the files with the output. The regression is obvious
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Updated] (JENA-256) Significant performance regression
(TDB?) on 2.7.1 RC compared to May 15 build
Posted by "Simon Helsen (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/JENA-256?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Simon Helsen updated JENA-256:
------------------------------
Attachment: (was: 271May15Output3.txt)
> Significant performance regression (TDB?) on 2.7.1 RC compared to May 15 build
> ------------------------------------------------------------------------------
>
> Key: JENA-256
> URL: https://issues.apache.org/jira/browse/JENA-256
> Project: Apache Jena
> Issue Type: Bug
> Components: TDB
> Affects Versions: Jena 2.7.1
> Environment: Windows 7, 64 bit, tested against 2.7.1 RC from June 9 versus a build from May 15
> Reporter: Simon Helsen
> Priority: Critical
> Attachments: 271May15Output1.txt, 271May15Output2.txt, 271May15Output3.txt, 271May15Output4.txt, 271RCOutput1.txt, 271RCOutput2.txt, 271RCOutput3.txt, 271RCOutput4.txt, PerformanceRegressionTest.java
>
>
> See also http://mail-archives.apache.org/mod_mbox/jena-dev/201206.mbox/%3C4FD9F19E.3080904%40bristol.ac.uk%3E
> I was able to reproduce the performance regression with an isolated test scenario. So I recreated the components ARQ, CORE, IRI, and TDB with the SVN state of May 15 9 (so svn update -r {2012-05-15} and then svn clean install)
> I then created a simple test program (attached as PerformanceRegressionTest.java) which I ran 4 times in a row for each version of Jena. Note that I deleted the TDB directory after the first 4 runs before using the other Jena version. Attached are the files with the output. The regression is obvious
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Updated] (JENA-256) Significant performance regression
(TDB?) on 2.7.1 RC compared to May 15 build
Posted by "Simon Helsen (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/JENA-256?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Simon Helsen updated JENA-256:
------------------------------
Attachment: PerformanceRegressionTest.java
> Significant performance regression (TDB?) on 2.7.1 RC compared to May 15 build
> ------------------------------------------------------------------------------
>
> Key: JENA-256
> URL: https://issues.apache.org/jira/browse/JENA-256
> Project: Apache Jena
> Issue Type: Bug
> Components: TDB
> Affects Versions: Jena 2.7.1
> Environment: Windows 7, 64 bit, tested against 2.7.1 RC from June 9 versus a build from May 15
> Reporter: Simon Helsen
> Priority: Critical
> Attachments: 271May15Output1.txt, 271May15Output2.txt, 271May15Output3.txt, 271May15Output4.txt, 271RCOutput1.txt, 271RCOutput2.txt, 271RCOutput3.txt, 271RCOutput4.txt, PerformanceRegressionTest.java
>
>
> See also http://mail-archives.apache.org/mod_mbox/jena-dev/201206.mbox/%3C4FD9F19E.3080904%40bristol.ac.uk%3E
> I was able to reproduce the performance regression with an isolated test scenario. So I recreated the components ARQ, CORE, IRI, and TDB with the SVN state of May 15 9 (so svn update -r {2012-05-15} and then svn clean install)
> I then created a simple test program (attached as PerformanceRegressionTest.java) which I ran 4 times in a row for each version of Jena. Note that I deleted the TDB directory after the first 4 runs before using the other Jena version. Attached are the files with the output. The regression is obvious
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (JENA-256) Significant performance regression
(TDB?) on 2.7.1 RC compared to May 15 build
Posted by "Simon Helsen (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/JENA-256?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13418601#comment-13418601 ]
Simon Helsen commented on JENA-256:
-----------------------------------
Strange, so I regenerated a build from HEAD (and checked for the batch size) and I still get similar performances as I mentioned before. (this is using Oracle 6). One thing is that before I start the test, I remove the index on disk. I suspect that is why the first run is faster. Also, I don't understand your last comment "it is worth trying your example code with a singe read transaction around the whole loop". If I do that, I create a nested transaction which is not permitted. I'll try with apache-jena-2.7.3-20120718.212752-16.zip as well
> Significant performance regression (TDB?) on 2.7.1 RC compared to May 15 build
> ------------------------------------------------------------------------------
>
> Key: JENA-256
> URL: https://issues.apache.org/jira/browse/JENA-256
> Project: Apache Jena
> Issue Type: Bug
> Components: TDB
> Affects Versions: Jena 2.7.1
> Environment: Windows 7, 64 bit, tested against 2.7.1 RC from June 9 versus a build from May 15
> Reporter: Simon Helsen
> Assignee: Andy Seaborne
> Priority: Critical
> Fix For: TDB 0.9.3
>
> Attachments: 271May15Output1.txt, 271May15Output2.txt, 271May15Output3.txt, 271May15Output4.txt, 271RCOutput1.txt, 271RCOutput2.txt, 271RCOutput3.txt, 271RCOutput4.txt, PerformanceRegressionTest.java
>
>
> See also http://mail-archives.apache.org/mod_mbox/jena-dev/201206.mbox/%3C4FD9F19E.3080904%40bristol.ac.uk%3E
> I was able to reproduce the performance regression with an isolated test scenario. So I recreated the components ARQ, CORE, IRI, and TDB with the SVN state of May 15 9 (so svn update -r {2012-05-15} and then svn clean install)
> I then created a simple test program (attached as PerformanceRegressionTest.java) which I ran 4 times in a row for each version of Jena. Note that I deleted the TDB directory after the first 4 runs before using the other Jena version. Attached are the files with the output. The regression is obvious
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (JENA-256) Significant performance regression
(TDB?) on 2.7.1 RC compared to May 15 build
Posted by "Simon Helsen (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/JENA-256?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13418287#comment-13418287 ]
Simon Helsen commented on JENA-256:
-----------------------------------
Hmmm, perhaps I was still running unoptimized. I used the following snapshot: https://repository.apache.org/content/repositories/snapshots/org/apache/jena/apache-jena/2.7.3-SNAPSHOT/apache-jena-2.7.3-20120718.050811-15.zip
Just wondering, why does the current code base have the unoptimized version by default? Is there a system property I can set to make sure I am using a non-0 batch size. Also, what was the batch size you used? Does it make a difference what size you use?
> Significant performance regression (TDB?) on 2.7.1 RC compared to May 15 build
> ------------------------------------------------------------------------------
>
> Key: JENA-256
> URL: https://issues.apache.org/jira/browse/JENA-256
> Project: Apache Jena
> Issue Type: Bug
> Components: TDB
> Affects Versions: Jena 2.7.1
> Environment: Windows 7, 64 bit, tested against 2.7.1 RC from June 9 versus a build from May 15
> Reporter: Simon Helsen
> Assignee: Andy Seaborne
> Priority: Critical
> Fix For: TDB 0.9.3
>
> Attachments: 271May15Output1.txt, 271May15Output2.txt, 271May15Output3.txt, 271May15Output4.txt, 271RCOutput1.txt, 271RCOutput2.txt, 271RCOutput3.txt, 271RCOutput4.txt, PerformanceRegressionTest.java
>
>
> See also http://mail-archives.apache.org/mod_mbox/jena-dev/201206.mbox/%3C4FD9F19E.3080904%40bristol.ac.uk%3E
> I was able to reproduce the performance regression with an isolated test scenario. So I recreated the components ARQ, CORE, IRI, and TDB with the SVN state of May 15 9 (so svn update -r {2012-05-15} and then svn clean install)
> I then created a simple test program (attached as PerformanceRegressionTest.java) which I ran 4 times in a row for each version of Jena. Note that I deleted the TDB directory after the first 4 runs before using the other Jena version. Attached are the files with the output. The regression is obvious
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (JENA-256) Significant performance regression
(TDB?) on 2.7.1 RC compared to May 15 build
Posted by "Simon Helsen (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/JENA-256?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13418609#comment-13418609 ]
Simon Helsen commented on JENA-256:
-----------------------------------
ok, so I tried the drop, but that didn't make a difference. Then I realized, looking at the code that I was probably running in mapped mode which, on Windows is shaky (and known to be slower), so I explicitly changed to use direct mode. The first run went fine (and fast), but the second run failed with the following exception (systematically):
Exception in thread "main" com.hp.hpl.jena.tdb.base.block.BlockException: BlockAccessBase: Bounds exception: C:\Temp\jenaTests\indexTestLocation\GSPO.dat: (40,40)
at com.hp.hpl.jena.tdb.base.file.BlockAccessBase.check(BlockAccessBase.java:107)
at com.hp.hpl.jena.tdb.base.file.BlockAccessBase.check(BlockAccessBase.java:114)
at com.hp.hpl.jena.tdb.base.file.BlockAccessDirect.write(BlockAccessDirect.java:85)
at com.hp.hpl.jena.tdb.base.file.BlockAccessDirect.overwrite(BlockAccessDirect.java:104)
at com.hp.hpl.jena.tdb.base.block.BlockMgrFileAccess.overwrite(BlockMgrFileAccess.java:102)
at com.hp.hpl.jena.tdb.base.block.BlockMgrWrapper.overwrite(BlockMgrWrapper.java:89)
at com.hp.hpl.jena.tdb.base.block.BlockMgrSync.overwrite(BlockMgrSync.java:90)
at com.hp.hpl.jena.tdb.base.block.BlockMgrCache.overwrite(BlockMgrCache.java:206)
at com.hp.hpl.jena.tdb.transaction.JournalControl.replay(JournalControl.java:305)
at com.hp.hpl.jena.tdb.transaction.JournalControl.recoverSegment(JournalControl.java:175)
at com.hp.hpl.jena.tdb.transaction.JournalControl.recoverFromJournal(JournalControl.java:128)
at com.hp.hpl.jena.tdb.StoreConnection._makeAndCache(StoreConnection.java:237)
at com.hp.hpl.jena.tdb.StoreConnection.make(StoreConnection.java:196)
at com.hp.hpl.jena.tdb.transaction.DatasetGraphTransaction.<init>(DatasetGraphTransaction.java:76)
at com.hp.hpl.jena.tdb.sys.TDBMakerTxn._create(TDBMakerTxn.java:49)
at com.hp.hpl.jena.tdb.sys.TDBMakerTxn.createDatasetGraph(TDBMakerTxn.java:37)
at com.hp.hpl.jena.tdb.TDBFactory._createDatasetGraph(TDBFactory.java:104)
at com.hp.hpl.jena.tdb.TDBFactory.createDatasetGraph(TDBFactory.java:73)
at com.hp.hpl.jena.tdb.TDBFactory.createDataset(TDBFactory.java:52)
at com.hp.hpl.jena.tdb.TDBFactory.createDataset(TDBFactory.java:48)
at com.ibm.jena.test.PerformanceRegressionTest.main(PerformanceRegressionTest.java:54)
It is reproducible and the only thing I did was add the following static block:
static {
ARQ.getContext().set(SystemTDB.symFileMode, "direct");
}
Can you reproduce this?
> Significant performance regression (TDB?) on 2.7.1 RC compared to May 15 build
> ------------------------------------------------------------------------------
>
> Key: JENA-256
> URL: https://issues.apache.org/jira/browse/JENA-256
> Project: Apache Jena
> Issue Type: Bug
> Components: TDB
> Affects Versions: Jena 2.7.1
> Environment: Windows 7, 64 bit, tested against 2.7.1 RC from June 9 versus a build from May 15
> Reporter: Simon Helsen
> Assignee: Andy Seaborne
> Priority: Critical
> Fix For: TDB 0.9.3
>
> Attachments: 271May15Output1.txt, 271May15Output2.txt, 271May15Output3.txt, 271May15Output4.txt, 271RCOutput1.txt, 271RCOutput2.txt, 271RCOutput3.txt, 271RCOutput4.txt, PerformanceRegressionTest.java
>
>
> See also http://mail-archives.apache.org/mod_mbox/jena-dev/201206.mbox/%3C4FD9F19E.3080904%40bristol.ac.uk%3E
> I was able to reproduce the performance regression with an isolated test scenario. So I recreated the components ARQ, CORE, IRI, and TDB with the SVN state of May 15 9 (so svn update -r {2012-05-15} and then svn clean install)
> I then created a simple test program (attached as PerformanceRegressionTest.java) which I ran 4 times in a row for each version of Jena. Note that I deleted the TDB directory after the first 4 runs before using the other Jena version. Attached are the files with the output. The regression is obvious
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (JENA-256) Significant performance regression
(TDB?) on 2.7.1 RC compared to May 15 build
Posted by "Andy Seaborne (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/JENA-256?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13418155#comment-13418155 ]
Andy Seaborne commented on JENA-256:
------------------------------------
I have run your example code with and without the write batching optimization:
(6111, 6144, 6125) vs (21320, 21815, 22079)
The optimized version is using the overnight build of July 18/19.
The unoptimized version is current code base with write batch size set to zero.
Setup: Ubuntu 12.04, Oracle java 6 and Oracle java 7.
With batched writes:
All 100 write operations wrote 5250 triples and took 7388ms
Total opening transaction time was 152ms
Total commit transaction time was 6111ms
All 100 write operations wrote 5250 triples and took 7392ms
Total opening transaction time was 150ms
Total commit transaction time was 6144ms
All 100 write operations wrote 5250 triples and took 7415ms
Total opening transaction time was 148ms
Total commit transaction time was 6125ms
Without batched writes:
All 100 write operations wrote 5250 triples and took 22869ms
Total opening transaction time was 184ms
Total commit transaction time was 21320ms
All 100 write operations wrote 5250 triples and took 23327ms
Total opening transaction time was 187ms
Total commit transaction time was 21815ms
All 100 write operations wrote 5250 triples and took 23581ms
Total opening transaction time was 196ms
Total commit transaction time was 22079ms
Also - the numbers of several runs are much more consistent that
> Significant performance regression (TDB?) on 2.7.1 RC compared to May 15 build
> ------------------------------------------------------------------------------
>
> Key: JENA-256
> URL: https://issues.apache.org/jira/browse/JENA-256
> Project: Apache Jena
> Issue Type: Bug
> Components: TDB
> Affects Versions: Jena 2.7.1
> Environment: Windows 7, 64 bit, tested against 2.7.1 RC from June 9 versus a build from May 15
> Reporter: Simon Helsen
> Assignee: Andy Seaborne
> Priority: Critical
> Fix For: TDB 0.9.3
>
> Attachments: 271May15Output1.txt, 271May15Output2.txt, 271May15Output3.txt, 271May15Output4.txt, 271RCOutput1.txt, 271RCOutput2.txt, 271RCOutput3.txt, 271RCOutput4.txt, PerformanceRegressionTest.java
>
>
> See also http://mail-archives.apache.org/mod_mbox/jena-dev/201206.mbox/%3C4FD9F19E.3080904%40bristol.ac.uk%3E
> I was able to reproduce the performance regression with an isolated test scenario. So I recreated the components ARQ, CORE, IRI, and TDB with the SVN state of May 15 9 (so svn update -r {2012-05-15} and then svn clean install)
> I then created a simple test program (attached as PerformanceRegressionTest.java) which I ran 4 times in a row for each version of Jena. Note that I deleted the TDB directory after the first 4 runs before using the other Jena version. Attached are the files with the output. The regression is obvious
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (JENA-256) Significant performance regression
(TDB?) on 2.7.1 RC compared to May 15 build
Posted by "Andy Seaborne (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/JENA-256?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13419024#comment-13419024 ]
Andy Seaborne commented on JENA-256:
------------------------------------
You can have separate non-nested transactions by doing:
Dataset dataset = TDBFactory.createDataset(location);
Dataset dataset1 = TDBFactory.createDataset(location);
and use one dataset as a reader and one as the writer. Just don't start two overlapping writes on two different datasets on the same thread as a write requests is queued until it can be granted.
> Significant performance regression (TDB?) on 2.7.1 RC compared to May 15 build
> ------------------------------------------------------------------------------
>
> Key: JENA-256
> URL: https://issues.apache.org/jira/browse/JENA-256
> Project: Apache Jena
> Issue Type: Bug
> Components: TDB
> Affects Versions: Jena 2.7.1
> Environment: Windows 7, 64 bit, tested against 2.7.1 RC from June 9 versus a build from May 15
> Reporter: Simon Helsen
> Assignee: Andy Seaborne
> Priority: Critical
> Fix For: TDB 0.9.3
>
> Attachments: 271May15Output1.txt, 271May15Output2.txt, 271May15Output3.txt, 271May15Output4.txt, 271RCOutput1.txt, 271RCOutput2.txt, 271RCOutput3.txt, 271RCOutput4.txt, PerformanceRegressionTest.java
>
>
> See also http://mail-archives.apache.org/mod_mbox/jena-dev/201206.mbox/%3C4FD9F19E.3080904%40bristol.ac.uk%3E
> I was able to reproduce the performance regression with an isolated test scenario. So I recreated the components ARQ, CORE, IRI, and TDB with the SVN state of May 15 9 (so svn update -r {2012-05-15} and then svn clean install)
> I then created a simple test program (attached as PerformanceRegressionTest.java) which I ran 4 times in a row for each version of Jena. Note that I deleted the TDB directory after the first 4 runs before using the other Jena version. Attached are the files with the output. The regression is obvious
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Updated] (JENA-256) Significant performance regression
(TDB?) on 2.7.1 RC compared to May 15 build
Posted by "Simon Helsen (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/JENA-256?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Simon Helsen updated JENA-256:
------------------------------
Attachment: (was: 271May15Output4.txt)
> Significant performance regression (TDB?) on 2.7.1 RC compared to May 15 build
> ------------------------------------------------------------------------------
>
> Key: JENA-256
> URL: https://issues.apache.org/jira/browse/JENA-256
> Project: Apache Jena
> Issue Type: Bug
> Components: TDB
> Affects Versions: Jena 2.7.1
> Environment: Windows 7, 64 bit, tested against 2.7.1 RC from June 9 versus a build from May 15
> Reporter: Simon Helsen
> Priority: Critical
> Attachments: 271May15Output1.txt, 271May15Output2.txt, 271May15Output3.txt, 271May15Output4.txt, 271RCOutput1.txt, 271RCOutput2.txt, 271RCOutput3.txt, 271RCOutput4.txt, PerformanceRegressionTest.java
>
>
> See also http://mail-archives.apache.org/mod_mbox/jena-dev/201206.mbox/%3C4FD9F19E.3080904%40bristol.ac.uk%3E
> I was able to reproduce the performance regression with an isolated test scenario. So I recreated the components ARQ, CORE, IRI, and TDB with the SVN state of May 15 9 (so svn update -r {2012-05-15} and then svn clean install)
> I then created a simple test program (attached as PerformanceRegressionTest.java) which I ran 4 times in a row for each version of Jena. Note that I deleted the TDB directory after the first 4 runs before using the other Jena version. Attached are the files with the output. The regression is obvious
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (JENA-256) Significant performance regression
(TDB?) on 2.7.1 RC compared to May 15 build
Posted by "Simon Helsen (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/JENA-256?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13295185#comment-13295185 ]
Simon Helsen commented on JENA-256:
-----------------------------------
And while this is probably expected, the time is lost in the commit. I am attaching an updated version which keeps the dataset creation outside and adds a few more numbers. Here are the summaries of 3 subsequent tests, first on the RC and then on the May 15 build (and all on Sun JRE 6):
RC - test 1:
All 100 write operations wrote 5250 triples and took 7548ms
Total opening transaction time was 380ms
Total commit transaction time was 5647ms
RC - test 2:
All 100 write operations wrote 5250 triples and took 12040ms
Total opening transaction time was 387ms
Total commit transaction time was 10081ms
RC - test 3:
All 100 write operations wrote 5250 triples and took 11761ms
Total opening transaction time was 377ms
Total commit transaction time was 10061ms
May 15 - test 1:
All 100 write operations wrote 5250 triples and took 4317ms
Total opening transaction time was 384ms
Total commit transaction time was 2853ms
May 15 - test 2:
All 100 write operations wrote 5250 triples and took 5410ms
Total opening transaction time was 399ms
Total commit transaction time was 3672ms
May 15 - test 3:
All 100 write operations wrote 5250 triples and took 5266ms
Total opening transaction time was 384ms
Total commit transaction time was 3405ms
> Significant performance regression (TDB?) on 2.7.1 RC compared to May 15 build
> ------------------------------------------------------------------------------
>
> Key: JENA-256
> URL: https://issues.apache.org/jira/browse/JENA-256
> Project: Apache Jena
> Issue Type: Bug
> Components: TDB
> Affects Versions: Jena 2.7.1
> Environment: Windows 7, 64 bit, tested against 2.7.1 RC from June 9 versus a build from May 15
> Reporter: Simon Helsen
> Priority: Critical
> Attachments: 271May15Output1.txt, 271May15Output2.txt, 271May15Output3.txt, 271May15Output4.txt, 271RCOutput1.txt, 271RCOutput2.txt, 271RCOutput3.txt, 271RCOutput4.txt, PerformanceRegressionTest.java
>
>
> See also http://mail-archives.apache.org/mod_mbox/jena-dev/201206.mbox/%3C4FD9F19E.3080904%40bristol.ac.uk%3E
> I was able to reproduce the performance regression with an isolated test scenario. So I recreated the components ARQ, CORE, IRI, and TDB with the SVN state of May 15 9 (so svn update -r {2012-05-15} and then svn clean install)
> I then created a simple test program (attached as PerformanceRegressionTest.java) which I ran 4 times in a row for each version of Jena. Note that I deleted the TDB directory after the first 4 runs before using the other Jena version. Attached are the files with the output. The regression is obvious
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (JENA-256) Significant performance regression
(TDB?) on 2.7.1 RC compared to May 15 build
Posted by "Andy Seaborne (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/JENA-256?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13428210#comment-13428210 ]
Andy Seaborne commented on JENA-256:
------------------------------------
This should be fixed now. The bug was not related to the transaction batching, that just happened to show it up.
> Significant performance regression (TDB?) on 2.7.1 RC compared to May 15 build
> ------------------------------------------------------------------------------
>
> Key: JENA-256
> URL: https://issues.apache.org/jira/browse/JENA-256
> Project: Apache Jena
> Issue Type: Bug
> Components: TDB
> Affects Versions: Jena 2.7.1
> Environment: Windows 7, 64 bit, tested against 2.7.1 RC from June 9 versus a build from May 15
> Reporter: Simon Helsen
> Assignee: Andy Seaborne
> Priority: Critical
> Fix For: TDB 0.9.3
>
> Attachments: 271May15Output1.txt, 271May15Output2.txt, 271May15Output3.txt, 271May15Output4.txt, 271RCOutput1.txt, 271RCOutput2.txt, 271RCOutput3.txt, 271RCOutput4.txt, PerformanceRegressionTest.java
>
>
> See also http://mail-archives.apache.org/mod_mbox/jena-dev/201206.mbox/%3C4FD9F19E.3080904%40bristol.ac.uk%3E
> I was able to reproduce the performance regression with an isolated test scenario. So I recreated the components ARQ, CORE, IRI, and TDB with the SVN state of May 15 9 (so svn update -r {2012-05-15} and then svn clean install)
> I then created a simple test program (attached as PerformanceRegressionTest.java) which I ran 4 times in a row for each version of Jena. Note that I deleted the TDB directory after the first 4 runs before using the other Jena version. Attached are the files with the output. The regression is obvious
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (JENA-256) Significant performance regression
(TDB?) on 2.7.1 RC compared to May 15 build
Posted by "Andy Seaborne (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/JENA-256?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13398580#comment-13398580 ]
Andy Seaborne commented on JENA-256:
------------------------------------
This is not a regression. This is effect of one of a number of fixes based on code analysis that make the system more robust to sudden death failures (external factors, e.g. lost of power or operating system level process kill (e.g. kill -9)). They are more likely to cause database corruption in direct mode where caching is separate from the filing system. That the failure modes have not been reported in the wild suggests that are rare (and not test for extensively).
Had the system been acting correctly in the first place, there would not be a change in performance.
Optimization of write-back can be done but, I believe, needs careful testing. The obvious technique is to batch write backs but then the system must be careful about (1) increased memory usage due to delays in write-back and (2) no active causing the queue not to be flushed. Crude prototyping by settign a write-back size of 10 transactions shows it works, using the test case, but issues (1) and (2) need understanding before putting the main codebase.
A partial improvement is to sync only files that have actually been updated but it seems the OS (Linux at least) is quite efficient here anyway.
Batching effects can also be seen by start a read transaction before the main loop, and ending it after the loop. In this case, the read blocks immediate write back and all non-journal changes happen at once at the end, but then there is a clear point to flush the delayed write-back queue.
Oddly, as the %-age of the time a read transaction is active increases under load, the performance may well improve because batchign effects emerge. This may explain why initial reports saw less of a change as they were read/write mixes, not singe threaded write only access patterns.
> Significant performance regression (TDB?) on 2.7.1 RC compared to May 15 build
> ------------------------------------------------------------------------------
>
> Key: JENA-256
> URL: https://issues.apache.org/jira/browse/JENA-256
> Project: Apache Jena
> Issue Type: Bug
> Components: TDB
> Affects Versions: Jena 2.7.1
> Environment: Windows 7, 64 bit, tested against 2.7.1 RC from June 9 versus a build from May 15
> Reporter: Simon Helsen
> Priority: Critical
> Attachments: 271May15Output1.txt, 271May15Output2.txt, 271May15Output3.txt, 271May15Output4.txt, 271RCOutput1.txt, 271RCOutput2.txt, 271RCOutput3.txt, 271RCOutput4.txt, PerformanceRegressionTest.java
>
>
> See also http://mail-archives.apache.org/mod_mbox/jena-dev/201206.mbox/%3C4FD9F19E.3080904%40bristol.ac.uk%3E
> I was able to reproduce the performance regression with an isolated test scenario. So I recreated the components ARQ, CORE, IRI, and TDB with the SVN state of May 15 9 (so svn update -r {2012-05-15} and then svn clean install)
> I then created a simple test program (attached as PerformanceRegressionTest.java) which I ran 4 times in a row for each version of Jena. Note that I deleted the TDB directory after the first 4 runs before using the other Jena version. Attached are the files with the output. The regression is obvious
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (JENA-256) Significant performance regression
(TDB?) on 2.7.1 RC compared to May 15 build
Posted by "Simon Helsen (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/JENA-256?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13417721#comment-13417721 ]
Simon Helsen commented on JENA-256:
-----------------------------------
Unfortunately, the new fixes don't seem to make a big difference (if anything made things a bit worse, compared to 2.7.1):
2.7.3 (July 18) - test 1:
All 100 write operations wrote 5250 triples and took 7771ms
Total opening transaction time was 235ms
Total commit transaction time was 6380ms
2.7.3 (July 18) - test 2:
All 100 write operations wrote 5250 triples and took 14248ms
Total opening transaction time was 203ms
Total commit transaction time was 12334ms
2.7.3 (July 18) - test 3:
All 100 write operations wrote 5250 triples and took 12404ms
Total opening transaction time was 183ms
Total commit transaction time was 10948ms
> Significant performance regression (TDB?) on 2.7.1 RC compared to May 15 build
> ------------------------------------------------------------------------------
>
> Key: JENA-256
> URL: https://issues.apache.org/jira/browse/JENA-256
> Project: Apache Jena
> Issue Type: Bug
> Components: TDB
> Affects Versions: Jena 2.7.1
> Environment: Windows 7, 64 bit, tested against 2.7.1 RC from June 9 versus a build from May 15
> Reporter: Simon Helsen
> Assignee: Andy Seaborne
> Priority: Critical
> Fix For: TDB 0.9.3
>
> Attachments: 271May15Output1.txt, 271May15Output2.txt, 271May15Output3.txt, 271May15Output4.txt, 271RCOutput1.txt, 271RCOutput2.txt, 271RCOutput3.txt, 271RCOutput4.txt, PerformanceRegressionTest.java
>
>
> See also http://mail-archives.apache.org/mod_mbox/jena-dev/201206.mbox/%3C4FD9F19E.3080904%40bristol.ac.uk%3E
> I was able to reproduce the performance regression with an isolated test scenario. So I recreated the components ARQ, CORE, IRI, and TDB with the SVN state of May 15 9 (so svn update -r {2012-05-15} and then svn clean install)
> I then created a simple test program (attached as PerformanceRegressionTest.java) which I ran 4 times in a row for each version of Jena. Note that I deleted the TDB directory after the first 4 runs before using the other Jena version. Attached are the files with the output. The regression is obvious
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (JENA-256) Significant performance regression
(TDB?) on 2.7.1 RC compared to May 15 build
Posted by "Andy Seaborne (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/JENA-256?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13295170#comment-13295170 ]
Andy Seaborne commented on JENA-256:
------------------------------------
Quick questions:
On what system were these executed? Windows vs Linux? JRE? etc etc.
Why does it do "TDBFactory.createDataset(location);" each time? If outside the loop, what happens?
> Significant performance regression (TDB?) on 2.7.1 RC compared to May 15 build
> ------------------------------------------------------------------------------
>
> Key: JENA-256
> URL: https://issues.apache.org/jira/browse/JENA-256
> Project: Apache Jena
> Issue Type: Bug
> Components: TDB
> Affects Versions: Jena 2.7.1
> Environment: Windows 7, 64 bit, tested against 2.7.1 RC from June 9 versus a build from May 15
> Reporter: Simon Helsen
> Priority: Critical
> Attachments: 271May15Output1.txt, 271May15Output2.txt, 271May15Output3.txt, 271May15Output4.txt, 271RCOutput1.txt, 271RCOutput2.txt, 271RCOutput3.txt, 271RCOutput4.txt, PerformanceRegressionTest.java
>
>
> See also http://mail-archives.apache.org/mod_mbox/jena-dev/201206.mbox/%3C4FD9F19E.3080904%40bristol.ac.uk%3E
> I was able to reproduce the performance regression with an isolated test scenario. So I recreated the components ARQ, CORE, IRI, and TDB with the SVN state of May 15 9 (so svn update -r {2012-05-15} and then svn clean install)
> I then created a simple test program (attached as PerformanceRegressionTest.java) which I ran 4 times in a row for each version of Jena. Note that I deleted the TDB directory after the first 4 runs before using the other Jena version. Attached are the files with the output. The regression is obvious
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Updated] (JENA-256) Significant performance regression
(TDB?) on 2.7.1 RC compared to May 15 build
Posted by "Simon Helsen (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/JENA-256?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Simon Helsen updated JENA-256:
------------------------------
Attachment: (was: 271May15Output2.txt)
> Significant performance regression (TDB?) on 2.7.1 RC compared to May 15 build
> ------------------------------------------------------------------------------
>
> Key: JENA-256
> URL: https://issues.apache.org/jira/browse/JENA-256
> Project: Apache Jena
> Issue Type: Bug
> Components: TDB
> Affects Versions: Jena 2.7.1
> Environment: Windows 7, 64 bit, tested against 2.7.1 RC from June 9 versus a build from May 15
> Reporter: Simon Helsen
> Priority: Critical
> Attachments: 271May15Output1.txt, 271May15Output2.txt, 271May15Output3.txt, 271May15Output4.txt, 271RCOutput1.txt, 271RCOutput2.txt, 271RCOutput3.txt, 271RCOutput4.txt, PerformanceRegressionTest.java
>
>
> See also http://mail-archives.apache.org/mod_mbox/jena-dev/201206.mbox/%3C4FD9F19E.3080904%40bristol.ac.uk%3E
> I was able to reproduce the performance regression with an isolated test scenario. So I recreated the components ARQ, CORE, IRI, and TDB with the SVN state of May 15 9 (so svn update -r {2012-05-15} and then svn clean install)
> I then created a simple test program (attached as PerformanceRegressionTest.java) which I ran 4 times in a row for each version of Jena. Note that I deleted the TDB directory after the first 4 runs before using the other Jena version. Attached are the files with the output. The regression is obvious
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (JENA-256) Significant performance regression
(TDB?) on 2.7.1 RC compared to May 15 build
Posted by "Simon Helsen (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/JENA-256?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13295173#comment-13295173 ]
Simon Helsen commented on JENA-256:
-----------------------------------
I wrote in the environment description of this issue that it was Windows 7/64 bit. I used IBM JRE 6, but I can try with SUN's as well if you want, but before you ask all these questions, why not simply try the test on your local system, once with the May 15 build and once with the RC? So far, 2 independent parties have reported the slowdown and you asked for a small reproducible test case. That is what I did.
The reason it does TDBFactory.createDataset(location) each time is that this is exactly how we write the code in our product because our product is designed to work multi-threaded (as per documentation). But yes, I can check what happens if the createDataset is outside the loop.
> Significant performance regression (TDB?) on 2.7.1 RC compared to May 15 build
> ------------------------------------------------------------------------------
>
> Key: JENA-256
> URL: https://issues.apache.org/jira/browse/JENA-256
> Project: Apache Jena
> Issue Type: Bug
> Components: TDB
> Affects Versions: Jena 2.7.1
> Environment: Windows 7, 64 bit, tested against 2.7.1 RC from June 9 versus a build from May 15
> Reporter: Simon Helsen
> Priority: Critical
> Attachments: 271May15Output1.txt, 271May15Output2.txt, 271May15Output3.txt, 271May15Output4.txt, 271RCOutput1.txt, 271RCOutput2.txt, 271RCOutput3.txt, 271RCOutput4.txt, PerformanceRegressionTest.java
>
>
> See also http://mail-archives.apache.org/mod_mbox/jena-dev/201206.mbox/%3C4FD9F19E.3080904%40bristol.ac.uk%3E
> I was able to reproduce the performance regression with an isolated test scenario. So I recreated the components ARQ, CORE, IRI, and TDB with the SVN state of May 15 9 (so svn update -r {2012-05-15} and then svn clean install)
> I then created a simple test program (attached as PerformanceRegressionTest.java) which I ran 4 times in a row for each version of Jena. Note that I deleted the TDB directory after the first 4 runs before using the other Jena version. Attached are the files with the output. The regression is obvious
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Closed] (JENA-256) Significant performance regression
(TDB?) on 2.7.1 RC compared to May 15 build
Posted by "Andy Seaborne (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/JENA-256?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Andy Seaborne closed JENA-256.
------------------------------
> Significant performance regression (TDB?) on 2.7.1 RC compared to May 15 build
> ------------------------------------------------------------------------------
>
> Key: JENA-256
> URL: https://issues.apache.org/jira/browse/JENA-256
> Project: Apache Jena
> Issue Type: Bug
> Components: TDB
> Affects Versions: Jena 2.7.1
> Environment: Windows 7, 64 bit, tested against 2.7.1 RC from June 9 versus a build from May 15
> Reporter: Simon Helsen
> Assignee: Andy Seaborne
> Priority: Critical
> Fix For: TDB 0.9.3
>
> Attachments: 271May15Output1.txt, 271May15Output2.txt, 271May15Output3.txt, 271May15Output4.txt, 271RCOutput1.txt, 271RCOutput2.txt, 271RCOutput3.txt, 271RCOutput4.txt, PerformanceRegressionTest.java
>
>
> See also http://mail-archives.apache.org/mod_mbox/jena-dev/201206.mbox/%3C4FD9F19E.3080904%40bristol.ac.uk%3E
> I was able to reproduce the performance regression with an isolated test scenario. So I recreated the components ARQ, CORE, IRI, and TDB with the SVN state of May 15 9 (so svn update -r {2012-05-15} and then svn clean install)
> I then created a simple test program (attached as PerformanceRegressionTest.java) which I ran 4 times in a row for each version of Jena. Note that I deleted the TDB directory after the first 4 runs before using the other Jena version. Attached are the files with the output. The regression is obvious
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira