[
https://issues.apache.org/jira/browse/IMPALA-12583?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17799172#comment-17799172
]
Wenzhe Zhou edited comment on IMPALA-12583 at 12/20/23 10:05 PM:
-----------------------------------------------------------------
There are following options to support "sys" and "information_schema" databases:
1. Access the tables of those two databases through Hive server with Hive jdbc
driver
Issue: It does not work in the deployment without Hive Server. It's not
efficient.
2. Access the tables directly from the Postgres (or MySQL) of HMS
We need to create "sys" and "information_schema" databases in Impala SQL syntax.
Issue: Keep schema in sync with Hive and handle schema upgrade in the future
3. Support the Hive JDBC Storage tables in Impala external data source
This assume the tables are handled by Impala built-in JDBC data source. We need
to covert the table configurations from Hive to Impala.
Issue: Need HMS to create the databases and tables if there is no Hive Server.
Work to do: Impala frontend to parse/analyze Hive JDBC table, read
configuration like jdbc url/username/password from hive-site.xml, support
built-in functions (restrict_information_schema(), current_authorizer(),
current_groups(), array_contains()) for access control.
4. Port the whole Hive JDBC Storage handler to Impala
Current choice is third option.
was (Author: wzhou):
There are following options to support "sys" and "information_schema" databases:
1. Access the tables of those two databases through Hive server with Hive jdbc
driver
Issue: It does not work in the deployment without Hive Server. It's not
efficient.
2. Access the tables directly from the Postgres (or MySQL) of HMS
We need to create "sys" and "information_schema" databases in Impala SQL syntax.
Issue: Keep schema in sync with Hive and handle schema upgrade in the future
3. Support the Hive JDBC Storage tables in Impala external data source
This assume the tables are handled by Impala built-in JDBC data source. We need
to covert the table configurations from Hive to Impala.
Issue: Need HMS to create the databases if there is no Hive Server.
Work to do: Impala frontend to parse/analyze Hive JDBC table, read
configuration like jdbc url/username/password from Hive-site.xml, support
built-in functions for access control.
4. Port the whole Hive JDBC Storage handler to Impala
Current choice is third option.
> Support reading hive "information_schema" views in Impala
> ---------------------------------------------------------
>
> Key: IMPALA-12583
> URL: https://issues.apache.org/jira/browse/IMPALA-12583
> Project: IMPALA
> Issue Type: Sub-task
> Reporter: Manish Maheshwari
> Assignee: Wenzhe Zhou
> Priority: Major
> Attachments: image-2023-11-30-02-24-18-869.png, information_schema.txt
>
>
> Hive supports "information_schema" db that all jdbc tables exposed from the
> HMS database. The same jdbc source tables should be queryable in Impala too.
>
> !image-2023-11-30-02-24-18-869.png!
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
