kafka-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Matthias J. Sax (JIRA)" <j...@apache.org>
Subject [jira] [Created] (KAFKA-6400) Consider setting default cache size to zero in Kafka Streams
Date Sat, 23 Dec 2017 20:37:02 GMT
Matthias J. Sax created KAFKA-6400:

             Summary: Consider setting default cache size to zero in Kafka Streams
                 Key: KAFKA-6400
                 URL: https://issues.apache.org/jira/browse/KAFKA-6400
             Project: Kafka
          Issue Type: Improvement
          Components: streams
    Affects Versions: 1.0.0
            Reporter: Matthias J. Sax
            Priority: Minor

Since the introduction of record caching in Kafka Streams DSL, we see regular reports/questions
of first times users about "Kafka Streams does not emit anything" or "Kafka Streams loses
messages". Those report are subject to record caching but no bugs and indicate bad user experience.

We might consider setting the default cache size to zero to avoid those issues and improve
the experience for first time users. This hold especially for simple word-count-demos (Note,
many people don't copy out example word-count but build their own first demo app.)

Remark: before we had caching, many users got confused about our update semantics and that
we emit an output record for each input record for windowed aggregation (ie, please give me
the "final" result"). Thus, we need to consider this and judge with care to not go "forth
and back" with default user experience -- we did have less questions about this behavior lately.

This message was sent by Atlassian JIRA

View raw message