[
https://issues.jboss.org/browse/TEIID-1598?page=com.atlassian.jira.plugin...
]
Steven Hawkins commented on TEIID-1598:
---------------------------------------
MetaMatrix used to function similar to that. It leads to good cache utilization when you
have a small set of fixed user queries. But then it's usually better just to rely on
user query result set caching. In most other scenarios materialization leads to better
cache usage.
That said there are scenarios that do benefit from caching translator queries especially
since we don't yet optimize common sub expressions with in a single user query (such
as issuing the same translator query multiple times).
Ttl can be configurable via translator property, but there's an issue of when to use
cached query results. In MMX caching needed to be enabled on the connector (translator)
and the query had to be cache eligible (non-update, non-transactional, with result set
caching enabled). That last condition meant that unless the source query was shared among
many user requests the user result set cache could supersede the source query cache.
translator scoped result set caching
------------------------------------
Key: TEIID-1598
URL:
https://issues.jboss.org/browse/TEIID-1598
Project: Teiid
Issue Type: Feature Request
Reporter: Mark Addleman
Assignee: Steven Hawkins
Priority: Minor
Something Steven mentioned in Teiid-1589 got me thinking: It would be helpful for
individual translators to request that Teiid cache results sets that match the query as
presented to the execution factory (after the planner is done rewriting the query based on
the translator's capabilities).
I envision a new method on an Execution getCacheTtlMs() where zero or negative indicates
that Teiid shouldn't cache the results.
--
This message is automatically generated by JIRA.
For more information on JIRA, see:
http://www.atlassian.com/software/jira