Github user ron8hu commented on a diff in the pull request:
https://github.com/apache/spark/pull/17138#discussion_r104760579
--- Diff:
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/optimizer/CostBasedJoinReorder.scala
---
@@ -0,0 +1,274 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.spark.sql.catalyst.optimizer
+
+import scala.collection.mutable
+
+import org.apache.spark.sql.catalyst.CatalystConf
+import org.apache.spark.sql.catalyst.expressions.{And, AttributeSet,
Expression, PredicateHelper}
+import org.apache.spark.sql.catalyst.plans.{Inner, InnerLike}
+import org.apache.spark.sql.catalyst.plans.logical.{Join, LogicalPlan,
Project}
+import org.apache.spark.sql.catalyst.rules.Rule
+
+
+/**
+ * Cost-based join reorder.
+ * We may have several join reorder algorithms in the future. This class
is the entry of these
+ * algorithms, and chooses which one to use.
+ */
+case class CostBasedJoinReorder(conf: CatalystConf) extends
Rule[LogicalPlan] with PredicateHelper {
+ def apply(plan: LogicalPlan): LogicalPlan = {
+ if (!conf.cboEnabled || !conf.joinReorderEnabled) {
+ plan
+ } else {
+ plan transform {
+ case p @ Project(projectList, j @ Join(_, _, _: InnerLike, _)) if
!j.ordered =>
+ reorder(j, p.outputSet)
+ case j @ Join(_, _, _: InnerLike, _) if !j.ordered =>
+ reorder(j, j.outputSet)
+ }
+ }
+ }
+
+ def reorder(plan: LogicalPlan, output: AttributeSet): LogicalPlan = {
+ val (items, conditions) = extractInnerJoins(plan)
+ val result =
+ if (items.size > 2 && items.size <= conf.joinReorderDPThreshold &&
conditions.nonEmpty) {
+ JoinReorderDP(conf, items, conditions,
output).search().getOrElse(plan)
+ } else {
+ plan
+ }
+ // Set all inside joins ordered.
+ setOrdered(result)
+ result
+ }
+
+ /**
+ * Extract inner joinable items and join conditions.
+ * This method works for bushy trees and left/right deep trees.
+ */
+ def extractInnerJoins(plan: LogicalPlan): (Seq[LogicalPlan],
Set[Expression]) = plan match {
+ case j @ Join(left, right, _: InnerLike, cond) =>
+ val (leftPlans, leftConditions) = extractInnerJoins(left)
+ val (rightPlans, rightConditions) = extractInnerJoins(right)
+ (leftPlans ++ rightPlans,
cond.map(splitConjunctivePredicates).getOrElse(Nil).toSet ++
+ leftConditions ++ rightConditions)
+ case Project(_, j @ Join(left, right, _: InnerLike, cond)) =>
+ val (leftPlans, leftConditions) = extractInnerJoins(left)
+ val (rightPlans, rightConditions) = extractInnerJoins(right)
+ (leftPlans ++ rightPlans,
cond.map(splitConjunctivePredicates).getOrElse(Nil).toSet ++
+ leftConditions ++ rightConditions)
+ case _ =>
+ (Seq(plan), Set())
+ }
+
+ def setOrdered(plan: LogicalPlan): Unit = plan match {
+ case j @ Join(left, right, _: InnerLike, cond) =>
+ j.ordered = true
+ setOrdered(left)
+ setOrdered(right)
+ case Project(_, j @ Join(left, right, _: InnerLike, cond)) =>
+ j.ordered = true
+ setOrdered(left)
+ setOrdered(right)
+ case _ =>
+ }
+}
+
+/**
+ * Reorder the joins using a dynamic programming algorithm:
--- End diff --
@hvanhovell We had a meeting with Sameer and Wenchen on 2/21/2017. We did
not meet you as you were not in San Francisco office on that day. In the
meeting, we agreed to have a good join reorder algorithm implemented in CBO's
first release as long as the algorithm has a good reference base. We can
improve the join reorder algorithm later in CBO's second release. After all,
we run short of time for Spark 2.2. We decided to use the algorithm in
Selinger's paper. For CBO's first release, we will be happy with the join
reorder algorithm if it can improve TPC-DS query performance without causing
regression.
---
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 [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
