[
https://issues.apache.org/jira/browse/FLINK-3230?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15247514#comment-15247514
]
ASF GitHub Bot commented on FLINK-3230:
---------------------------------------
Github user uce commented on the pull request:
https://github.com/apache/flink/pull/1910#issuecomment-211847621
Great addition to have a producer to Kinesis. The changes look good
overall, I had some minor inline comments.
The main concern is that we can only test this with a real Kinesis setup,
which makes it fragile for future changes etc. Did you consider trying to mock
the `KinesisProducer` to test that the expected calls are issued to it?
PS: The title of the JIRA issue and this PR is confusing (the last `...
into Flink` part)
> Kinesis streaming producer
> --------------------------
>
> Key: FLINK-3230
> URL: https://issues.apache.org/jira/browse/FLINK-3230
> Project: Flink
> Issue Type: Sub-task
> Components: Streaming Connectors
> Reporter: Tzu-Li (Gordon) Tai
> Assignee: Robert Metzger
>
> Add a FlinkKinesisProducer for the Flink Kinesis streaming connector. We will
> be using AWS SDK implementation for code consistency with the
> FlinkKinesisConsumer.
> The features of FlinkKinesisProducer is rather straightforward:
> 1. Partition put records based on partition key.
> 2. Configurable put mode: Bulk put for higher throughput vs. sequential
> single record puts. Size of bulk should also be configurable.
> 3. For bulk put, user can also choose to enforce strict ordering of the
> result with the tradeoff of higher put latency. Ref:
> https://brandur.org/kinesis-order
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
