[jira] [Commented] (SENTRY-2030) Enable first-level cache for DataNucleus

["Brian Towles (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/SENTRY-2030?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16227267#comment-16227267 ] 

Brian Towles commented on SENTRY-2030:
--------------------------------------

I think thats reasonable.  My main concern would be synchronization in an HA deployment.  But honestly I dont feel that nodes being different by up to a minute or so is a major issue, As long as a reasonable TTL is put in place for the cache.

What about making it a configurable option that defaults to enabled?

> Enable first-level cache for DataNucleus
> ----------------------------------------
>
>                 Key: SENTRY-2030
>                 URL: https://issues.apache.org/jira/browse/SENTRY-2030
>             Project: Sentry
>          Issue Type: Bug
>          Components: Sentry
>    Affects Versions: 2.0.0
>            Reporter: Alexander Kolbasov
>
> Sentry is using a new persistence manager instance per transaction and repeatable-read transaction isolation level (except for Oracle). I think we can safely enable first-level caching (which is at the PersistenceManager level) which should improve performance without introducing any issues.
> [~spena] [~kkalyan] [~btowles][~lina.li] What do you think?



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)