GitHub user sachouche opened a pull request:
https://github.com/apache/drill/pull/1060
DRILL-5846: Improve parquet performance for Flat Data Types
Performance improvements for the Parquet Scanner (Flat Data Types). The are
two flags to control this performance enhancement (disabled by default):
Option I -
Config Name: store.parquet.flat.reader.bulk
Config Type : boolean
Description : Enables bulk processing to minimize memory checks and
improve JVM HotSpot optimizations
Option II -
Config Name: scan.optimized.implicit.columns
Config Type : boolean
Description : Memory optimization when storing duplicate value (implicit
columns have duplicate values within a batch). Code profiling indicated this
step represented one third of Parquet processing (when implicit columns are
processed).
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/sachouche/drill drill-parquet-improv
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/drill/pull/1060.patch
To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:
This closes #1060
----
commit 0a3a8b053be85c570ff24237d4737f37668383bd
Author: Salim Achouche <[email protected]>
Date: 2017-12-04T01:53:08Z
DRILL-5846: Improve parquet performance for Flat Data Types
----
---
