Github user fhueske commented on the issue:
https://github.com/apache/flink/pull/3423
Hi @shaoxuan-wang, sorry. I missed your comment. If you haven't started
reworking the batch aggregations yet, I agree to do it later and just change
the merging to smaller batches.
So far, we always put an emphasis on robustness and tried to avoid memory
issues as much as possible. In batch, most of the JVM memory is maintained by
Flink and not available for regular user-function objects. So I'd suggest to be
a bit conservative here and batch 16 rows. We can also run a few benchmarks to
check how much the parameter affects the performance.
---
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.
---
