Details
-
New Feature
-
Status: Open
-
Major
-
Resolution: Unresolved
-
0.6.0
-
None
-
None
Description
The DefaultChooser currently defaults to a RoundRobinChooser when no chooser is specified in config. This means that, by default, we just round robin over which stream's messages to process next.
It might be possible to come up with a better strategy than round robin in this scenario. This problem is outlined in detail in:
https://wiki.apache.org/samza/Pluggable%20MessageChooser
The most interesting and intuitive strategy seems to be to try and maintain the alignment of the streams at the time the job starts (the fourth strategy outlined in the wiki proposal).
Also, see SAMZA-2 for more discussion on this.