[
https://issues.apache.org/jira/browse/HIVE-11611?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14707433#comment-14707433
]
Sergio Peña commented on HIVE-11611:
------------------------------------
[~Ferd]. Thanks for working on this.
[~rdblue] I see you fixed the validation on PARQUET-363. If we now pass an
empty schema to Parquet, how is 'select count(1)' going to count the # of
records if Parquet does not return any data? I am not sure how Parquet handles
an empty schema, could you help me understand this? Do you think that
requesting at least 1 column to Parquet will allow Hive to count the records
(See [~Ferd] patch)?
> A bad performance regression issue with Parquet happens if Hive does not
> select any columns
> -------------------------------------------------------------------------------------------
>
> Key: HIVE-11611
> URL: https://issues.apache.org/jira/browse/HIVE-11611
> Project: Hive
> Issue Type: Sub-task
> Affects Versions: 2.0.0
> Reporter: Sergio Peña
> Assignee: Ferdinand Xu
> Attachments: HIVE-11611.patch
>
>
> A possible performance issue may happen with the below code when using a
> query like this {{SELECT count(1) FROM parquetTable}}.
> {code}
> if (!ColumnProjectionUtils.isReadAllColumns(configuration) &&
> !indexColumnsWanted.isEmpty()) {
> MessageType requestedSchemaByUser =
> getSchemaByIndex(tableSchema, columnNamesList,
> indexColumnsWanted);
> return new ReadContext(requestedSchemaByUser, contextMetadata);
> } else {
> return new ReadContext(tableSchema, contextMetadata);
> }
> {code}
> If there are not columns nor indexes selected, then the above code will read
> the full schema from Parquet even if Hive does not do anything with such
> values.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
