You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ignite.apache.org by "Konstantin Orlov (Jira)" <ji...@apache.org> on 2024/01/19 07:59:00 UTC
[jira] [Created] (IGNITE-21311) Sql. Partition pruning. Introduce pruning for correlated scans
Konstantin Orlov created IGNITE-21311:
-----------------------------------------
Summary: Sql. Partition pruning. Introduce pruning for correlated scans
Key: IGNITE-21311
URL: https://issues.apache.org/jira/browse/IGNITE-21311
Project: Ignite
Issue Type: Improvement
Components: sql
Reporter: Konstantin Orlov
Apart of static pruning made on preparation step, we may also introduce dynamic pruning for correlated scans.
In case of correlated join, a predicate on the right shoulder should be evaluated based on the context provided from the left shoulder. This context is not known until runtime, those partitions are not trimmed on preparation step. However, this very case doesn't differ match from the static pruning: the only difference here is a time when pruning should be applied.
In order to support dynamic pruning, we should save pruning meta during planning phase in a scan node. Later, in runtime, we should evaluate pruning function in order to derive partitions satisfying the predicate. Those partitions should be used to do a lookup into a table.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)