[
https://issues.apache.org/jira/browse/LENS-376?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14358713#comment-14358713
]
Amareshwari Sriramadasu commented on LENS-376:
----------------------------------------------
Here are the options that I could think of :
# JDBCDriver uses CubeMetastoreClient.getTable(), which already caches all hive
tables
** Will be introducing CubeMetastoreClient dependency on JDBCDriver. Right now,
Driver does not use anything from lens-cube module. But will be added going
forward.
# JDBCDriver itself builds its own cache of tables over time.
** No dependency with lens-cube
** But adds more memory and takes time to build the cache.
I'm inclined towards option #1. Thoughts?
> ColumnarSQLRewriter always ends up making metastore lookup
> ----------------------------------------------------------
>
> Key: LENS-376
> URL: https://issues.apache.org/jira/browse/LENS-376
> Project: Apache Lens
> Issue Type: Improvement
> Components: driver-jdbc, server
> Reporter: Jaideep Dhok
> Fix For: 2.1
>
>
> ColumnarSQLRewriter does a metastore lookup to check if table names and db
> names need to be replaced. This introduces network calls in the estimate path.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
