[
https://issues.apache.org/jira/browse/OAK-6747?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Julian Reschke resolved OAK-6747.
---------------------------------
Resolution: Later
Ran VersionGCTest with scale=1000 for three DBs, twice, with and without the
change.
trunk:
{noformat}
DB2: 2.149 min
DB2: 2.179 min
Oracle: 1.328 min
Oracle: 1.846 min
PostgreSQL: 1.627 min
PostgreSQL: 1.391 min
{noformat}
with patch:
{noformat}
DB2: 2.042 min
DB2: 1.928 min
Oracle: 1.887 min
Oracle: 1.501 min
PostgreSQL: 1.824 min
PostgreSQL: 1.848 min
{noformat}
Needs more research.
> improve index handling in RDBVersionGCSupport
> ---------------------------------------------
>
> Key: OAK-6747
> URL: https://issues.apache.org/jira/browse/OAK-6747
> Project: Jackrabbit Oak
> Issue Type: Improvement
> Components: rdbmk
> Reporter: Julian Reschke
> Assignee: Julian Reschke
> Attachments: OAK-6747.diff
>
>
> We currently find split documents by using "LIKE" to exclude regular
> documents. However, these "LIKE" expressions start with a wildcard, which
> defeats use of the index.
> We may be able to improve this by adding to the already existing exclusion
> patterns, such as:
> {noformat}
> 0:/% 1:/% 2:/% 3:/% 4:/% ...
> {noformat}
> We should benchmark this first, though. See OAK-6719.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
