[
https://issues.apache.org/jira/browse/SPARK-26345?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17307590#comment-17307590
]
Li Xian commented on SPARK-26345:
---------------------------------
[~yumwang] I think the current implementation has a problem. the pages returned
by `readNextFilteredRowGroup` may not be aligned, some columns may have more
rows than others.
Parquet is using `org.apache.parquet.column.impl.SynchronizingColumnReader`
with `rowIndexes` to make sure that rows are aligned.
Currently `VectorizedParquetRecordReader` doesn't have such synchronizing among
pages from different columns. Using `readNextFilteredRowGroup` may result in
incorrect result.
> Parquet support Column indexes
> ------------------------------
>
> Key: SPARK-26345
> URL: https://issues.apache.org/jira/browse/SPARK-26345
> Project: Spark
> Issue Type: Umbrella
> Components: SQL
> Affects Versions: 3.1.0
> Reporter: Yuming Wang
> Assignee: Yuming Wang
> Priority: Major
> Fix For: 3.2.0
>
>
> Parquet 1.11 supports column indexing. Spark can supports this feature for
> better read performance.
> More details:
> https://issues.apache.org/jira/browse/PARQUET-1201
>
> Benchmark result:
> [https://github.com/apache/spark/pull/31393#issuecomment-769767724]
> This feature is enabled by default, and users can disable it by setting
> {{parquet.filter.columnindex.enabled}} to false.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
