You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@drill.apache.org by "Arina Ielchiieva (Jira)" <ji...@apache.org> on 2020/03/27 13:07:00 UTC

[jira] [Created] (DRILL-7672) Make metadata type required when reading from / writing into Drill Metastore

Arina Ielchiieva created DRILL-7672:
---------------------------------------

             Summary: Make metadata type required when reading from / writing into Drill Metastore
                 Key: DRILL-7672
                 URL: https://issues.apache.org/jira/browse/DRILL-7672
             Project: Apache Drill
          Issue Type: Improvement
    Affects Versions: 1.17.0
            Reporter: Arina Ielchiieva
            Assignee: Arina Ielchiieva
             Fix For: 1.18.0


Metastore consists of components: TABLES, VIEWS etc (so far only TABLES are implemented). Each component metadata can have types. For examples, TABLES metadata can be of the following types: TABLE, SEGMENT, FILE, ROW_GROUP, PARTITION.
During initial Metastore implementation when reading from / writing into Metastore, metadata type was indicated in filter expressions. 
For Iceberg Metastore where all data is stored in files this was not this critical, basically when information is retrieved about the table, table folder is queried.
For other Metastore implementations knowing metadata type can be more critical. For example, RDBMS Metastore would store TABLES metadata in different tables thus knowing which table to query would improve performance rather than trying to query all tables.
Of course, we could traverse query filter and look for the hints which metadata type is needed but it is much better to know required metadata type beforehand without any extra logic.
Taking into account that Metastore metadata is queried only in Drill code, developer knows beforehand what he needs to get / update / delete.

This Jira aims to make metadata type required when reading from / writing into Drill Metastore. This change does not have any affect on the users, just internal code refactoring.
 



--
This message was sent by Atlassian Jira
(v8.3.4#803005)