[ http://issues.apache.org/jira/browse/DERBY-1073?page=all ]
A B updated DERBY-1073:
-----------------------
Attachment: d1073_timeoutFix_v1.patch
As described in the derby-dev thread here:
http://article.gmane.org/gmane.comp.apache.db.derby.devel/16723
the original patch for DERBY-1073 was incomplete. For details see the above
thread. In short, the 'default' behavior for DERBY-1073 (i.e. the case where
there are no pushed predicates) combined with the changes for DERBY-1007 can
result in significantly greater compilation time for some queries--as evidenced
by the 4 to 7 time slow-down in the regression tests for nist/dml132.sql.
The resolution to this problem is to take the changes for DERBY-1073 one step
further by recognizing when a subquery's optimizer has timed out and, for each
round thereafter, allowing the optimizer to continue until it finds it's first
complete (and valid) plan for that round. Before d1073_timeoutFix_v1.patch the
optimizer for a subquery would return immediately (i.e. without doing any
optimization) for every round after the initial timeout, which led to incorrect
(and exceedingly high) cost estimates.
More simply put, the changes in d1073_timeoutFix_v1.patch delay subquery
timeout until the optimizer for the subquery has found a legitimate cost to
return.
In order to have a legitimate cost, the optimizer must find at least one
complete join order. Instead of just using the first possible join order, the
changes in d1073_timeoutFix_v1.patch try to guess at what the best join order
will be by "jumping" to the join order that was most recently deemed "best" for
the subquery prior to timeout. It then stops optimizing (times out) and
returns the cost of that join order.
Note that with d1073_timeoutFix_v1.patch, the nist/dml132.sql test will still
run more slowly than it did prior to DERBY-1007 and DERBY-1073. But a) the
slow-down will be far less than the 4 to 7-time slow-down seen in the current
trunk, and b) the slow-down is actually the result of more correct behavior.
The reason this slow-down is more correct is because it comes from the fact
that optimizer timeout value is now correct (roughly 35ish seconds) whereas
prior to DERBY-1007 and DERBY-1073 the timeout value, which was only 14-ish
seconds, was incorrect because the subquery that timed out was returning an
illegitimate cost estimate that in turn led to an incorrect timeout value.
I ran derbyall with these changes against sane jars on Red Hat with IBM 1.4.2
and saw no new failures.
I would appreciate any review comments, if anyone has the time...Thanks.
> Reset optimizer timeout for subqueries on a per-round basis to allow
> consideration of plans that use pushed predicates.
> -----------------------------------------------------------------------------------------------------------------------
>
> Key: DERBY-1073
> URL: http://issues.apache.org/jira/browse/DERBY-1073
> Project: Derby
> Type: Sub-task
> Components: Performance
> Versions: 10.2.0.0
> Reporter: A B
> Assignee: A B
> Priority: Minor
> Attachments: d1073_timeoutFix_v1.patch, d1073_v1.patch, d1073_v2.patch
>
> I wanted to file this as subtask to DERBY-805, but since DERBY-805 is itself
> a subtask to DERBY-649, I was not able to do so. So I'm creating this issue
> as a(nother) subtaks for DERBY-649.
> [ Based on derby-dev thread found here:
> http://article.gmane.org/gmane.comp.apache.db.derby.devel/16007 ]
> If we have a query such as:
> select <...> from
> (select t1.i, t2.j from t1, t2 where <...>) X1,
> T3
> where <...>
> then we would have one "outer" query and one "subquery". The outer query
> would be "select <...> from X1, T3", the subquery would be "select t1.i, t2.j
> from t1, t2".
> In this case the Derby optimizer will create two instances of OptimizerImpl:
> one for the outer query (call it OI_OQ) and one for the subquery (call it
> OI_SQ). Each OptimizerImpl has its own timeout "clock" that it initializes
> at creation time--but never resets. If timeout occurs, the OptimizerImpl
> will stop searching for "the" best plan and will just take the best plan
> found so far.
> That said, for every permutation of the outer query a call will be made to
> optimize the subquery. To simplify things, let's assume there are only two
> permutations of the outer query: one with join order {X1, T3} and another
> with join order {T3, X1}.
> Now let's say we're looking at the first permutation {X1, T3}. OI_OQ will
> make a call to optimize the subquery represented by OI_SQ. Let's further say
> that the subquery tries some permutation {T1, T2} and then times out. It
> then returns the plan information for {T1, T2} to the outer query. The outer
> query, which has *not* yet timed out, then decides to try its second
> permutation {T3, X1}. So it again makes a call to optimize the subquery. In
> this case, the subquery--which has already timed out--will *immediately*
> return without trying to optimize anything. The outer query will then make a
> decision about its second permutation based on the un-optimized subquery's
> plan results.
> This hasn't really been an issue to date because the "best plan" chosen by
> the subquery is typically independent of the outer query's current
> permutation--with the exception of "outerCost", which is passed in from the
> outer query and is factored into the subquery's cost estimates. Because of
> this relative independence, the plan chosen by the subquery would rarely (if
> ever?) change with different permutations of the outer query, so if the
> subquery timed out once there was no point in trying to re-optimize it again
> later.
> With DERBY-805, though, Derby has acquired the ability to push predicates
> from outer queries down into subqueries--which means that the outer join
> order can have a very significant impact on the plan chosen by the subquery.
> But because the timeout mechanism is never reset, we could end up skipping
> the second optimization phase of the subquery, which means we never get a
> chance to see how much the outer predicates can help, and thus we could end
> up skipping over some plans that have the potential to give us significant
> performance improvement.
> So resolution of this issue would involve resetting the timeout state for
> subqueries to allow the Derby optimizer to consider plans that rely on pushed
> predicates.
--
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
