Github user anisnasir commented on the pull request:
https://github.com/apache/flink/pull/1069#issuecomment-136496310
@tillrohrmann Yes, we can design a system that adapts to the load imbalance
and increases the number of partitions per key in real time. However, this
comes with few other questions that are:
1. When to increases the choices?
2. How much to increase the choices?
3. Should we increase the choices for all the keys?
A simple solution for the datasets that cannot be handled using our
solution is to use ShufflePartitioner.
---
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.
---
