You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@hive.apache.org by "Thai Bui (JIRA)" <ji...@apache.org> on 2018/10/26 18:57:00 UTC
[jira] [Commented] (HIVE-10459) Add materialized views to Hive
[ https://issues.apache.org/jira/browse/HIVE-10459?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16665550#comment-16665550 ]
Thai Bui commented on HIVE-10459:
---------------------------------
[~jcamachorodriguez] [~alangates] Would it be possible to support materialized views on external tables? For examples, there are cases where the tables are mostly immutable, or dropped & recreated as external tables with date table_name__yyyy_mm_dd postfix, with a stable view on top of that such tables as table_name -> table_name__yyyy_mm_dd, materialized views could work with cache invalidations and query result caching.
I'm currently hacking Hive to support such use cases since my company would mostly rely on immutable datasets on AWS S3 (by removing the transactional hardcoded constraints). However, I think Hive should support this feature as a first class citizen since the industry is moving towards more workloads in the cloud.
The feature in https://issues.apache.org/jira/browse/HIVE-19154 could be used to invalidate cache & push notifications externally into Hive via the cloud blob storage's change events as well. Although I am not familiar with the internal of Hive enough to understand how everything could work together.
> Add materialized views to Hive
> ------------------------------
>
> Key: HIVE-10459
> URL: https://issues.apache.org/jira/browse/HIVE-10459
> Project: Hive
> Issue Type: New Feature
> Components: Views
> Reporter: Alan Gates
> Assignee: Jesus Camacho Rodriguez
> Priority: Major
>
> Materialized views are useful as ways to store either alternate versions of data (e.g. same data, different sort order) or derivatives of data sets (e.g. commonly used aggregates). It is useful to store these as materialized views rather than as tables because it can give the optimizer the ability to understand how data sets are related.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)