[teiid-issues] [JBoss JIRA] (TEIID-1880) Support UNION ALL in multi source models

[Steven Hawkins (Commented) (JIRA)]

    [ https://issues.jboss.org/browse/TEIID-1880?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12654203#comment-12654203 ] 

Steven Hawkins commented on TEIID-1880:
---------------------------------------

Tim,

The multi-source planning logic isn't quite the same as normal planning.  The plan is first made against a single source assuming capabilities that are limited by what multisource supports - which doesn't include order by since that operation must logically be applied after unioning sources.  Then the plan is modified to add in various sources needed via a union.  This may result in a sub-optimial plan.  Partitioned views were added to be an explicit replacement for multi-source mode in 7.4, which is fully optimized.  This would require you to add a view for each "multi-source" table using a partitioning column, e.g. select a, b, 'uk' as source_name from x union all select c, d, 'us' from y
The additional benefit of this mode is that the metadata from each source will be fully represented, and it is not required that all table/column names must match across the sources.  The downside of course being the creation of the view layer. 

At some point we will look to bridge the gap between these two approaches, such as with TEIIDDES-1198, or by dynamically creating the needed views under the covers in Teiid.
                
> Support UNION ALL in multi source models
> ----------------------------------------
>
>                 Key: TEIID-1880
>                 URL: https://issues.jboss.org/browse/TEIID-1880
>             Project: Teiid
>          Issue Type: Feature Request
>            Reporter: Mark Addleman
>            Assignee: Steven Hawkins
>
> We have physical data sources that can have many millions of rows.  We are currently using Teiid's multi-source feature to union them together but, I believe Teiid is doing a proper UNION of the tables as opposed to a UNION ALL.  The implied distinct operation in the UNION slows down our queries a lot because many operations cannot be pushed down to the underlying data sources and must be resolved in Teiid.
> We would like the option of performing a UNION ALL on multi source models

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.jboss.org/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira