Github user davies commented on the pull request:
https://github.com/apache/spark/pull/6875#issuecomment-117438111
@lianhuiwang Aggregation is different than join, because aggregation could
aggregation could reduce the data size, but join cannot. Optimizer could figure
out whether use broadcast join or sort merge join based on the size of table,
but it's very hard to guess what's the memory assumption will for aggregation
(which is determined by the number of unique groups and aggregation algorithm).
All the aggregations happens within a partition, so no shuffling is needed.
Usually, there are two aggregations happen before and after shuffling.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
