You are viewing a plain text version of this content. The canonical link for it is here.
Posted to oak-issues@jackrabbit.apache.org by "Stefan Eissing (JIRA)" <ji...@apache.org> on 2017/03/20 09:24:41 UTC
[jira] [Commented] (OAK-5944) VersionGarbageCollector: allow to
update _deletedOnce in batch operations
[ https://issues.apache.org/jira/browse/OAK-5944?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15932372#comment-15932372 ]
Stefan Eissing commented on OAK-5944:
-------------------------------------
My measurements (dev laptop, SSD, local 3.4 mongo), 100GB aem database
{{oak-run revisions collect}}:
- previous version:
{code}
2565 docs/s (25229600 docs, 9835598.537 ms)
{code}
- this patch:
{code}
3946 docs/s (27818380 docs, 7049613.160 ms)
{code}
> VersionGarbageCollector: allow to update _deletedOnce in batch operations
> -------------------------------------------------------------------------
>
> Key: OAK-5944
> URL: https://issues.apache.org/jira/browse/OAK-5944
> Project: Jackrabbit Oak
> Issue Type: Improvement
> Components: documentmk
> Reporter: Julian Reschke
> Attachments: OAK-5944.diff
>
>
> Right now, we collect the update operations in batches, but then issue individual update operations.
> A special-case API that takes a set of updates and applies them in a single DB operation might speed up things, in particular when there's a network hop.
> Related to this, we may want to think about whether we need to attach a modified check to each single operation (by keeping the {{_modified}} value for each {{_id}}). Wouldn't it be sufficient to just check that the {{_modified}} value is not newer than the end of the time window for the VGC run?
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)