Github user ioana-delaney commented on a diff in the pull request:
https://github.com/apache/spark/pull/15363#discussion_r106255224
--- Diff:
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/optimizer/CostBasedJoinReorder.scala
---
@@ -51,6 +51,11 @@ case class CostBasedJoinReorder(conf: CatalystConf)
extends Rule[LogicalPlan] wi
def reorder(plan: LogicalPlan, output: AttributeSet): LogicalPlan = {
val (items, conditions) = extractInnerJoins(plan)
+ // Find the star schema joins. Currently, it returns the star join
with the largest
+ // fact table. In the future, it can return more than one star join
(e.g. F1-D1-D2
+ // and F2-D3-D4).
+ val starJoinPlans = StarSchemaDetection(conf).findStarJoins(items,
conditions.toSeq)
--- End diff --
@ron8hu One way to overcome the search space problem of a dynamic
programming join enumeration is to support various levels of optimization
algorithms. For example, the lowest level would employ a âgreedyâ algorithm
that would choose the âcheapestâ join that can be made over the remaining
sub-plans in each iteration. Then, we move up to higher levels of optimizations
such as dynamic programming join enumeration with various heuristics. Some good
heuristics that can be applied are: push Cartesian products planning towards
the end; produce only left-deep tree; apply star schema heuristics, etc. The
next level of optimization would be full, maximal optimization that considers
all possible join orders and spares no effort⦠Catalyst/CBO can start with
one level optimization, and automatically drop to âgreedyâ if it exceeds
the allocated search space. We can also give control to the user by
externalizing these levels of optimizations. If their query compile time is too
hi
gh, they can lower the level.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
