[
https://issues.apache.org/jira/browse/BEAM-2185?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16005711#comment-16005711
]
Raghu Angadi commented on BEAM-2185:
------------------------------------
Some more considerations :
- Splits:
- Each partition should be split further.
- If we could fetch the size in bytes, it is easier to split. I don't
think it is feasible get byte offsets.
- Split could be based on message offsets, but the size could be too large
or too small depending on average message size.
- Take a hint from the user if the size is not known? May be not. Read a
few sample records? Probably an over kill.
- It should support liquid-sharding (dynamic splitting). This does not need
average message size. Just the offsets are good enough.
> KafkaIO bounded source
> ----------------------
>
> Key: BEAM-2185
> URL: https://issues.apache.org/jira/browse/BEAM-2185
> Project: Beam
> Issue Type: New Feature
> Components: sdk-java-extensions
> Reporter: Raghu Angadi
>
> KafkaIO could be a useful source for batch applications as well. It could
> implement a bounded source. The primary question is how the bounds are
> specified.
> One option : Source specifies a time period (say 9am-10am), and KafkaIO
> fetches appropriate start and end offsets based on time-index in Kafka. This
> would suite many batch applications that are launched on a scheduled.
> Another option is to always read till the end and commit the offsets to
> Kafka. Handling failures and multiple runs of a task might be complicated.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
