[
https://issues.apache.org/jira/browse/HADOOP-5060?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12676065#action_12676065
]
Ari Rabkin commented on HADOOP-5060:
------------------------------------
There's substantial interest at Berkeley in doing some sort of in-collector
aggregation, to get very short latencies. It would be a big help if the code
written for this JIRA were modular in such a way that it could be pulled out of
the map-reduce framework, and run separately. It's already pretty
straightforward to extract Records in the collector.
> Create a generic aggregator for Chukwa
> --------------------------------------
>
> Key: HADOOP-5060
> URL: https://issues.apache.org/jira/browse/HADOOP-5060
> Project: Hadoop Core
> Issue Type: New Feature
> Components: contrib/chukwa
> Reporter: Jerome Boulon
> Assignee: Jerome Boulon
>
> Create a generic way to compute aggregation on top of chukwaRecords based on
> a config file
> Should be able:
> - work on several Chukwa streams
> - To aggregate by time period
> - Group by values for specific keys
> - Provide a redefine list of functions (AVG,MIN,MAX,Counter->Rate
> conversion...)
> - work with new functions
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
