[
https://issues.apache.org/jira/browse/SAMZA-334?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14063684#comment-14063684
]
Chris Riccomini commented on SAMZA-334:
---------------------------------------
bq. I am not sure what guarantees provides YARN for one to choose the machine
where a new scaled out Task will run.
Unfortunately, YARN is pretty bad at this right now (See SAMZA-335 for
discussion). We run the FairShare scheduler internally, which is a bit better,
since it tries to evenly distribute (round robin) containers among all nodes on
the grid. Still, it's not perfect.
> Need for asymmetric container config
> ------------------------------------
>
> Key: SAMZA-334
> URL: https://issues.apache.org/jira/browse/SAMZA-334
> Project: Samza
> Issue Type: Improvement
> Components: container
> Affects Versions: 0.8.0
> Reporter: Chinmay Soman
>
> The current (and upcoming) partitioning scheme(s) suggest that there might be
> a skew in the amount of data ingested and computation performed across
> different containers for a given Samza job. This directly affects the amount
> of resources required by a container - which today are completely symmetric.
> Case A] Partitioning on Kafka partitions
> For instance, consider a partitioner job which reads data from different
> Kafka topics (having different partition layouts). In this case, its possible
> that a lot of topics have a smaller number of Kafka partitions. Consequently
> the containers processing these partitions would need more resources than
> those responsible for the higher numbered partitions.
> Case B] Partitioning based on Kafka topics
> Even in this case, its very easy for some containers to be doing more work
> than others - leading to a skew in resource requirements.
> Today, the container config is based on the requirements for the worst (doing
> the most work) container. Needless to say, this leads to resource wastage. A
> better approach needs to consider what is the true requirement per container
> (instead of per job).
--
This message was sent by Atlassian JIRA
(v6.2#6252)
