db-derby-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "A B (JIRA)" <derby-...@db.apache.org>
Subject [jira] Created: (DERBY-1906) Investigate appropriateness of current optimizer timeout mechanism.
Date Fri, 29 Sep 2006 21:21:19 GMT
Investigate appropriateness of current optimizer timeout mechanism.

                 Key: DERBY-1906
                 URL: http://issues.apache.org/jira/browse/DERBY-1906
             Project: Derby
          Issue Type: Task
          Components: Performance, SQL
    Affects Versions:,,
            Reporter: A B

The Derby optimizer timeout mechanism is based on the assumption that cost estimates calculated
by the optimizer represent the number of milliseconds the optimizer thinks it will take for
the query to execute.  In order for this to be effective the cost estimates need to be relatively
accurate--or else, as described in DERBY-1905, the timeout mechanism becomes pretty ineffective
pretty quickly.

While it is true that incorrect cost estimates are perhaps the biggest cause of inefficient
optimizer timeout, I can't help but wonder if there might be some other way to handle optimizer
timeout that is not so dependent on the correctness of a costEstimate-to-milliseconds mapping.
 Maybe, for example, an algorithm based on delta's between query plan cost estimates would
be more appropriate?

I do not know what alternatives exist or which, if any, would be more "right" for Derby than
a millisecond-based approach, but I think it's worth searching out and/or coming up with algorithms/ideas
that might prove more effective for Derby optimizer timeout--especially when dealing with
very deeply-nested queries that have large FROM lists, such as those attached to DERBY-1777.

This message is automatically generated by JIRA.
If you think it was sent incorrectly contact one of the administrators: http://issues.apache.org/jira/secure/Administrators.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira


View raw message