[
https://issues.apache.org/jira/browse/OPENJPA-1469?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12803300#action_12803300
]
Kevin Sutter commented on OPENJPA-1469:
---------------------------------------
More test results...
DataCache = true
QueryCache = false (proposed new default)
RemoteCommitProvider = sjvm
30 Failures, 37 Errors
DataCache = true
QueryCache = true (current default)
RemoteCommitProvider = sjvm
394 Failures, 39 Errors
Two things are becoming obvious...
1) I don't have the time to resolve all of the errors associated with turning
on the DataCache by default before the 2.0 cutoff.
We can continue to investigate these errors and get this default reset in the
near future. Most of these problems are probably just testcase errors (like
the other fixups). But, we need to take the time to investigate and understand
before making this switch.
2) We have to turn the QueryCache off by default. Not only does the "true"
setting affect the performance gain when you turn on DataCache, but from a
quick peek at the errors generated by the above configuration, the use of the
QueryCache is affecting other aspects of the runtime like Serialization.
I will be creating separate sub-tasks for the work items not being covered.
Please comment if you do not agree with this direction.
Thanks,
Kevin
> Turn on DataCache by default
> ----------------------------
>
> Key: OPENJPA-1469
> URL: https://issues.apache.org/jira/browse/OPENJPA-1469
> Project: OpenJPA
> Issue Type: Improvement
> Components: datacache
> Affects Versions: 2.0.0-M1, 2.0.0-M2, 2.0.0-M3, 2.0.0-M4
> Reporter: Kevin Sutter
> Assignee: Kevin Sutter
> Fix For: 2.0.0
>
>
> I'm proposing that we change the default behavior for the DataCache.
> Currently, the DataCache is turned off by default. Our competition is
> turning on their L2 cache by default. Thus, out-of-the-box comparisons are
> killing us. We have to explain the difference, get them to change the
> configuration one way or the other for apples-to-apples comparisons, and then
> re-run the benchmark. By that time, it's too late. We already lost.
> Due to the upcoming 2.0.0 release, it seems like a good time to change this
> default. Please comment directly to this JIRA.
> Thanks,
> Kevin
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
