[
https://issues.apache.org/jira/browse/IGNITE-24131?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Maksim Timonin updated IGNITE-24131:
------------------------------------
Description: In case SQL query with group by, HashAggregateNode handles
every row and store in the hash-key group. It now creates object arr for every
row but it is sub-optimal for queries that operates big data sets (100k rows)
and only few groups. In a such case the array can be reused. (was: In case SQL
query with group by, by every row creates an array GroupKey just to find
whether this group already exists. In case queries operates big amounts of rows
and only few groups, this becomes non-optimal.)
> GroupKey.Builder should not create array for every row
> -------------------------------------------------------
>
> Key: IGNITE-24131
> URL: https://issues.apache.org/jira/browse/IGNITE-24131
> Project: Ignite
> Issue Type: Improvement
> Reporter: Maksim Timonin
> Assignee: Maksim Timonin
> Priority: Major
> Fix For: 2.18
>
> Time Spent: 10m
> Remaining Estimate: 0h
>
> In case SQL query with group by, HashAggregateNode handles every row and
> store in the hash-key group. It now creates object arr for every row but it
> is sub-optimal for queries that operates big data sets (100k rows) and only
> few groups. In a such case the array can be reused.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
