[
https://issues.apache.org/jira/browse/OAK-2984?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14583624#comment-14583624
]
Chetan Mehrotra commented on OAK-2984:
--------------------------------------
[~nleite] Would this also provide better concurrent write throughput. So far my
understanding was that Mongo 2.6.x had a db level locking and having collection
in separate database would help. However test done as part of OAK-1153 did not
show much difference. May be testcase were implemented incorrectly. So wanted
to confirm if there are other benefit apart from different mount point?
> Set blobs collection on separate database
> -----------------------------------------
>
> Key: OAK-2984
> URL: https://issues.apache.org/jira/browse/OAK-2984
> Project: Jackrabbit Oak
> Issue Type: Improvement
> Components: mongomk
> Reporter: Norberto Leite
> Labels: performance
>
> MongoDB allows the creation of several different database namespaces to allow
> concept separation, between different applications and separate workloads to
> perform conceptual and physical isolation.
> It also allows different databases to be set on [different mount points |
> http://docs.mongodb.org/manual/reference/configuration-options/#storage.directoryPerDB]
> which for this particular workload (nodes vs blobs) would be greatly
> appreciated. Having different physical disks would be to avoid IO contention
> provoked by demanding large blob operations upload operations on disk level.
> The implementation details would be as simple has defining an option that
> would require a *blobs database name*
> N.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
