[
https://issues.apache.org/jira/browse/FLINK-37481?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Gustavo de Morais updated FLINK-37481:
--------------------------------------
Description:
We currently always rely on a chain of binary joins when there are multiple
joins in a query. This can often generate a lot of intermediate state which
considerably increases the state size - specially for regular joins. There are
optimizations that can be done to a multi-way join operator that can optimize
the worst case scenarios for binary joins.
We want as a first step with this ticket to have a multi join operator that
works on multiple inputs for regular joins.
was:
We currently always rely on a chain of binary joins when there are multiple
joins in a query. There are optimizations that can be done to a multi way join
operator that can optimize the worst case scenarios for binary joins.
We want as a first step with this ticket to have a multi join operator that
works on multiple inputs.
> Multi way join operator
> -----------------------
>
> Key: FLINK-37481
> URL: https://issues.apache.org/jira/browse/FLINK-37481
> Project: Flink
> Issue Type: New Feature
> Reporter: Gustavo de Morais
> Priority: Major
> Labels: pull-request-available
>
> We currently always rely on a chain of binary joins when there are multiple
> joins in a query. This can often generate a lot of intermediate state which
> considerably increases the state size - specially for regular joins. There
> are optimizations that can be done to a multi-way join operator that can
> optimize the worst case scenarios for binary joins.
>
> We want as a first step with this ticket to have a multi join operator that
> works on multiple inputs for regular joins.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
