Github user haohui commented on the pull request:
https://github.com/apache/storm/pull/796#issuecomment-148118238
The main motivation is performance -- The code that generated by Calcite
has a lot of down casts which can potentially have a visible impact, as the
targeted throughput is millions of messages per seconds.
Bringing LLVM allows StormSQL to leverage the mature optimization and opens
up further opportunities including automatic vectorization, etc.
Note that LLVM is used only for optimization -- the next step is to
translate LLVM IR back to Java Bytecode (STORM-1097). The approach is quite
similar to using Janino to generate Java bytecode. There is no native
components in the worker node.
---
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.
---
