[
https://issues.apache.org/jira/browse/JCR-3828?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
BELUGA BEHR updated JCR-3828:
-----------------------------
Description:
The following implementation features several key changes:
1) Buffer records in memory and discard the buffer if the file is already in
the Data Store. This allows for very fast processing of duplicate files with
small overhead for files that do not collide.
2) Use of read/write locks so that multiple records can be submitted
simultaneously without blocking, but will be blocked if there are deletions
occurring.
3) Less synchronization in general.
> BufferedDataStore Implementation
> --------------------------------
>
> Key: JCR-3828
> URL: https://issues.apache.org/jira/browse/JCR-3828
> Project: Jackrabbit Content Repository
> Issue Type: Improvement
> Components: jackrabbit-data
> Reporter: BELUGA BEHR
> Priority: Trivial
> Attachments: BufferedDataStore.java, FileDataStoreMarkWalker.java
>
>
> The following implementation features several key changes:
> 1) Buffer records in memory and discard the buffer if the file is already in
> the Data Store. This allows for very fast processing of duplicate files with
> small overhead for files that do not collide.
> 2) Use of read/write locks so that multiple records can be submitted
> simultaneously without blocking, but will be blocked if there are deletions
> occurring.
> 3) Less synchronization in general.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
