You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@spark.apache.org by "Rajesh Balamohan (Jira)" <ji...@apache.org> on 2020/07/08 08:56:00 UTC
[jira] [Updated] (SPARK-32225) Parquet footer information is read
twice
[ https://issues.apache.org/jira/browse/SPARK-32225?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Rajesh Balamohan updated SPARK-32225:
-------------------------------------
Description:
When running queries, spark reads parquet footer information twice. In cloud env, this would turn out to be expensive (depending on the jobs, # of splits). It would be nice to reuse the footer information already read via "ParquetInputFormat::buildReaderWithPartitionValues"
!spark_parquet_footer_reads.png|width=640,height=644!
Lines of interest:
[https://github.com/apache/spark/blob/master/sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/parquet/ParquetFileFormat.scala#L271]
[https://github.com/apache/spark/blob/master/sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/parquet/ParquetFileFormat.scala#L326]
[https://github.com/apache/spark/blob/master/sql/core/src/main/java/org/apache/spark/sql/execution/datasources/parquet/SpecificParquetRecordReaderBase.java#L105]
[https://github.com/apache/spark/blob/master/sql/core/src/main/java/org/apache/spark/sql/execution/datasources/parquet/SpecificParquetRecordReaderBase.java#L111]
was:
When running queries, spark reads parquet footer information twice. In cloud env, this would turn out to be expensive (depending on the jobs, # of splits). It would be nice to reuse the footer information already read via "ParquetInputFormat::buildReaderWithPartitionValues"
!image-2020-07-08-14-24-23-470.png|width=726,height=730!
Lines of interest:
[https://github.com/apache/spark/blob/master/sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/parquet/ParquetFileFormat.scala#L271]
[https://github.com/apache/spark/blob/master/sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/parquet/ParquetFileFormat.scala#L326]
[https://github.com/apache/spark/blob/master/sql/core/src/main/java/org/apache/spark/sql/execution/datasources/parquet/SpecificParquetRecordReaderBase.java#L105]
[https://github.com/apache/spark/blob/master/sql/core/src/main/java/org/apache/spark/sql/execution/datasources/parquet/SpecificParquetRecordReaderBase.java#L111]
> Parquet footer information is read twice
> ----------------------------------------
>
> Key: SPARK-32225
> URL: https://issues.apache.org/jira/browse/SPARK-32225
> Project: Spark
> Issue Type: Improvement
> Components: SQL
> Affects Versions: 3.0.0
> Reporter: Rajesh Balamohan
> Priority: Minor
> Attachments: spark_parquet_footer_reads.png
>
>
> When running queries, spark reads parquet footer information twice. In cloud env, this would turn out to be expensive (depending on the jobs, # of splits). It would be nice to reuse the footer information already read via "ParquetInputFormat::buildReaderWithPartitionValues"
>
> !spark_parquet_footer_reads.png|width=640,height=644!
> Lines of interest:
> [https://github.com/apache/spark/blob/master/sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/parquet/ParquetFileFormat.scala#L271]
> [https://github.com/apache/spark/blob/master/sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/parquet/ParquetFileFormat.scala#L326]
>
> [https://github.com/apache/spark/blob/master/sql/core/src/main/java/org/apache/spark/sql/execution/datasources/parquet/SpecificParquetRecordReaderBase.java#L105]
> [https://github.com/apache/spark/blob/master/sql/core/src/main/java/org/apache/spark/sql/execution/datasources/parquet/SpecificParquetRecordReaderBase.java#L111]
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org