[
https://issues.apache.org/jira/browse/HIVE-15129?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Rajesh Balamohan updated HIVE-15129:
------------------------------------
Description:
When multiple queries are run in LLAP, stripe metadata cache misses were
observed even though enough memory was available.
https://github.com/apache/hive/blob/master/llap-server/src/java/org/apache/hadoop/hive/llap/io/encoded/OrcEncodedDataReader.java#L655.
Even in cases when data was found in cache, it wasn't getting used as
{{globalnc}} changed from query to query. This
would be lot more beneficial in cloud storage where opening and reading small
of data can be expensive compared to HDFS. Creating a superset of existing
indexes with {{globalInc}} would be helpful.
was:
When multiple queries are run in LLAP, stripe metadata cache misses were
observed even though enough memory was available.
https://github.com/apache/hive/blob/master/llap-server/src/java/org/apache/hadoop/hive/llap/io/encoded/OrcEncodedDataReader.java#L655.
Even though data would be found in cache, it wasn't getting used as
{{globalnc}} changed from query to query. This
would be lot more beneficial in cloud storage where opening and reading small
of data can be expensive compared to HDFS. Creating a superset of existing
indexes with {{globalInc}} would be helpful.
> LLAP : Enhance cache hits for stripe metadata across queries
> ------------------------------------------------------------
>
> Key: HIVE-15129
> URL: https://issues.apache.org/jira/browse/HIVE-15129
> Project: Hive
> Issue Type: Improvement
> Components: llap
> Reporter: Rajesh Balamohan
> Priority: Minor
> Attachments: HIVE-15129.1.patch
>
>
> When multiple queries are run in LLAP, stripe metadata cache misses were
> observed even though enough memory was available.
> https://github.com/apache/hive/blob/master/llap-server/src/java/org/apache/hadoop/hive/llap/io/encoded/OrcEncodedDataReader.java#L655.
> Even in cases when data was found in cache, it wasn't getting used as
> {{globalnc}} changed from query to query. This
> would be lot more beneficial in cloud storage where opening and reading small
> of data can be expensive compared to HDFS. Creating a superset of existing
> indexes with {{globalInc}} would be helpful.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
