You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@nifi.apache.org by "Mark Payne (JIRA)" <ji...@apache.org> on 2016/01/25 16:42:39 UTC
[jira] [Commented] (NIFI-1281) Investigate performance when queues
are swapping heavily
[ https://issues.apache.org/jira/browse/NIFI-1281?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15115419#comment-15115419 ]
Mark Payne commented on NIFI-1281:
----------------------------------
I looked into this some and i think that what we outlined above is accurate. i do believe it can be improved, but will take some refactoring to do so cleanly. Probably not something we can get done for 0.5.0 so I moved it to 0.6.0
> Investigate performance when queues are swapping heavily
> --------------------------------------------------------
>
> Key: NIFI-1281
> URL: https://issues.apache.org/jira/browse/NIFI-1281
> Project: Apache NiFi
> Issue Type: Task
> Components: Core Framework
> Reporter: Mark Payne
> Fix For: 0.6.0
>
>
> If we have a queue that has a huge amount of swapping going on (in and out), the ability to pull records from that queue is poor. When running a test, we had a producer continually pushing records into a queue with no backpressure. The consumer was very fast until swapping started to occur. At that point, the ability to consume dropped to only a few thousand events/second. Stopping the producer provided much better performance on the consuming side, which is problematic.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)