Github user drcrallen commented on the pull request:
https://github.com/apache/spark/pull/9243#issuecomment-150581844
General FYI, we did a lot of analysis on various bitmap implementations for
the Druid project. We were more interested in the direct-memory (mmap'd file)
capabilities. The general discussion thread can be found [in the dev
group](https://groups.google.com/d/msg/druid-development/_kw2jncIlp0/FLEhI36b7W0J)
The general findings were that Roaring (at the time) took up more space.
This has been much improved in the 0.5.x line. So an alternate solution to this
PR and the issue may be to upgrade to 0.5.x
Roaring on the 0.4.x line was generally comparable or slightly faster than
Concise in many aspects on our particular datasets in benchmarks against
production-like workloads (judged by query time). Roaring was way faster than
Concise at doing unions and intersections. Full analysis has not been re-run
since the 0.5.x release.
---
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]
