[
https://issues.apache.org/jira/browse/IMPALA-12253?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Yida Wu updated IMPALA-12253:
-----------------------------
Epic Link: IMPALA-13187 (was: IMPALA-11461)
> Improve the integration of codegen cache and async codegen
> ----------------------------------------------------------
>
> Key: IMPALA-12253
> URL: https://issues.apache.org/jira/browse/IMPALA-12253
> Project: IMPALA
> Issue Type: Improvement
> Components: Backend
> Reporter: Yida Wu
> Assignee: Yida Wu
> Priority: Major
>
> IMPALA-11470 adds support for codegen caching, from the single node perf A-B
> testing, codegen caching could perform better than async codegen when the
> cache hits for low latency queries, while the async codegen is faster for the
> first runs. To better integrate the codegen caching with the existing async
> codegen, we may optimize the workflow like below while async codegen and
> codegen caching both enabled:
> # Generate the codegen caching key before doing async codegen.
> # If key is existing in the cache, go codegen caching.
> # Otherwise, go async codegen.
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
