Mailing-List: contact dev-help@kafka.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@kafka.apache.org
Date: Thu, 30 Nov 2017 01:05:43 +0000 (UTC)
From: "Guozhang Wang (JIRA)" <jira@apache.org>
To: dev@kafka.apache.org
Message-ID: <JIRA.13121795.1512003636000.351431.1512003943212@Atlassian.JIRA>
In-Reply-To: <JIRA.13121795.1512003636000@Atlassian.JIRA>
References: <JIRA.13121795.1512003636000@Atlassian.JIRA> <JIRA.13121795.1512003636226@jira-lw-us.apache.org>
Subject: [jira] [Created] (KAFKA-6286) Consider reusing materialized store
 for multi-same-stream join
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit
archived-at: Thu, 30 Nov 2017 01:05:53 -0000

Guozhang Wang created KAFKA-6286:
------------------------------------

             Summary: Consider reusing materialized store for multi-same-stream join
                 Key: KAFKA-6286
                 URL: https://issues.apache.org/jira/browse/KAFKA-6286
             Project: Kafka
          Issue Type: Bug
          Components: streams
            Reporter: Guozhang Wang


Imagine the following streams application:

{code}
stream1.join(stream2...)..join(stream2)
{code}

Each join will result in {{stream2}} being materialized into a separate store. Arguably such multi-joins that involves the same stream multiple times is rare, but it worth considering if we can optimize such cases.

One thing to note, though, is that in our DSL parser today we do "put into store first, and then query the other store second", which means if we share the same store it would result in duplicates as the matching would already see the newly put records in the second join.


--
This message was sent by Atlassian JIRA
(v6.4.14#64029)