[
https://issues.redhat.com/browse/TEIID-5928?page=com.atlassian.jira.plugi...
]
Steven Hawkins commented on TEIID-5928:
---------------------------------------
After looking this over some more, the initial approach would be to add generic logic for
an additional load strategy. That is additional teiid_rel:MATVIEW_ options to control a
partitioned load. This would be something like adding:
teiid_rel:MATVIEW_PART_LOAD_COLUMN - name of the view column the load is partitioned over
teiid_rel:MATVIEW_PART_LOAD_VALUES - specifies the partitioning values, likely would
default to select distinct(load_column) from matView
I'll update SYS.Schemas to include the source names. So in the case of a multisource
materialization you would set the teiid_rel:MATVIEW_PART_LOAD_COLUMN to the relevant
column, possibly SOURCE_NAME and teiid_rel:MATVIEW_PART_LOAD_VALUES to
select cast(col as string) from (exec sys.arrayIterate((select sources from sys.schemas
where name = 'some schema'))) as a
I realize this isn't as friendly as Teiid automatically inferring those values, but
it ensures that we'll handle all cases initially without the need for detection logic
for the various cases multi-source vs. partitioned union, the effects of view layers, etc.
Allow External and Internal materialization of multi-source to
fail/update individually
---------------------------------------------------------------------------------------
Key: TEIID-5928
URL:
https://issues.redhat.com/browse/TEIID-5928
Project: Teiid
Issue Type: Enhancement
Components: Common
Affects Versions: 12.2.2
Reporter: Rafael Sampaio
Assignee: Steven Hawkins
Priority: Major
Fix For: 15.0
Original Estimate: 0 minutes
Time Spent: 30 minutes
Remaining Estimate: 0 minutes
Hi, all.
First of all thanks in advance for your attention, and efforts on such a great product.
Since I haven't found any docs or examples regarding my current usage (please if this
already exists or there's a better way of doing this, disregard this request and
advice) here goes my suggestion:
It would be nice to have a way to gracefully update/fail per source, if a view that has
Materialization enabled and target a multi-source source model, because things can get,
messy if I have to create a view per source and Materialize each one and only then create
a unified (UNION) view.
Using this approach also obligates me to implement custom insert/update/delete/based on
the discriminator column.
Thanks again.
--
This message was sent by Atlassian Jira
(v7.13.8#713008)