[
https://issues.apache.org/jira/browse/DRILL-6381?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16621505#comment-16621505
]
ASF GitHub Bot commented on DRILL-6381:
---------------------------------------
amansinha100 commented on a change in pull request #1466: DRILL-6381: Add
support for index based planning and execution
URL: https://github.com/apache/drill/pull/1466#discussion_r219032715
##########
File path:
exec/java-exec/src/main/java/org/apache/drill/exec/planner/physical/RowKeyJoinPrel.java
##########
@@ -0,0 +1,117 @@
+/*
+ * 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.drill.exec.planner.physical;
+
+import java.io.IOException;
+import java.util.Iterator;
+
+import org.apache.drill.exec.physical.base.PhysicalOperator;
+import org.apache.drill.exec.physical.config.RowKeyJoinPOP;
+import org.apache.drill.exec.planner.cost.DrillCostBase.DrillCostFactory;
+import org.apache.drill.exec.planner.physical.visitor.PrelVisitor;
+import org.apache.drill.exec.record.BatchSchema.SelectionVectorMode;
+import org.apache.calcite.rel.InvalidRelException;
+import org.apache.calcite.rel.RelNode;
+import org.apache.calcite.rel.core.Join;
+import org.apache.calcite.rel.core.JoinRelType;
+import org.apache.calcite.rel.metadata.RelMetadataQuery;
+import org.apache.calcite.rex.RexNode;
+import org.apache.calcite.plan.RelOptCluster;
+import org.apache.calcite.plan.RelOptCost;
+import org.apache.calcite.plan.RelOptPlanner;
+import org.apache.calcite.plan.RelTraitSet;
+
+import org.apache.drill.shaded.guava.com.google.common.base.Preconditions;
+
+public class RowKeyJoinPrel extends JoinPrel implements Prel {
+
+ double estimatedRowCount = -1;
+ public RowKeyJoinPrel(RelOptCluster cluster, RelTraitSet traits, RelNode
left, RelNode right,
+ RexNode condition, JoinRelType joinType) throws InvalidRelException {
+ super(cluster, traits, left, right, condition, joinType);
+ Preconditions.checkArgument(joinType == JoinRelType.INNER);
+ }
+
+ @Override
+ public PhysicalOperator getPhysicalOperator(PhysicalPlanCreator creator)
throws IOException {
+ PhysicalOperator leftPop = ((Prel)left).getPhysicalOperator(creator);
+ PhysicalOperator rightPop = ((Prel)right).getPhysicalOperator(creator);
+ RowKeyJoinPOP rkPop = new RowKeyJoinPOP(leftPop, rightPop);
+ return creator.addMetadata(this, rkPop);
+ }
+
+ @Override
+ public double estimateRowCount(RelMetadataQuery mq) {
+ if (estimatedRowCount >= 0) {
+ return estimatedRowCount;
+ }
+ return this.getLeft().getRows();
+ }
+
+ @Override
+ public Join copy(RelTraitSet traitSet, RexNode conditionExpr, RelNode left,
RelNode right,
+ JoinRelType joinType, boolean semiJoinDone) {
+ try {
+ RowKeyJoinPrel rkj = new RowKeyJoinPrel(this.getCluster(), traitSet,
left, right, conditionExpr, joinType);
+ rkj.setEstimatedRowCount(this.estimatedRowCount);
+ return rkj;
+ } catch (InvalidRelException e) {
+ throw new AssertionError(e);
+ }
+ }
+
+ @Override
+ public RelOptCost computeSelfCost(RelOptPlanner planner, RelMetadataQuery
mq) {
+ if(PrelUtil.getSettings(getCluster()).useDefaultCosting()) {
+ return super.computeSelfCost(planner).multiplyBy(.1);
+ }
+ double rowCount = mq.getRowCount(this.getRight());
+ DrillCostFactory costFactory = (DrillCostFactory) planner.getCostFactory();
+ return costFactory.makeCost(rowCount, 0, 0, 0,
Review comment:
RowKeyJoin operator by itself incurs negligible CPU and I/O cost since it is
not doing a real join. The actual cost is attributed to the skip-scan (random
I/O). The RK join will hold 1 batch in memory but since it is not a buffering
operator, we typically don't assign memory costs to them. I can add the
explanation.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
> Add capability to do index based planning and execution
> -------------------------------------------------------
>
> Key: DRILL-6381
> URL: https://issues.apache.org/jira/browse/DRILL-6381
> Project: Apache Drill
> Issue Type: New Feature
> Components: Execution - Relational Operators, Query Planning &
> Optimization
> Reporter: Aman Sinha
> Assignee: Aman Sinha
> Priority: Major
> Fix For: 1.15.0
>
>
> If the underlying data source supports indexes (primary and secondary
> indexes), Drill should leverage those during planning and execution in order
> to improve query performance.
> On the planning side, Drill planner should be enhanced to provide an
> abstraction layer which express the index metadata and statistics. Further,
> a cost-based index selection is needed to decide which index(es) are
> suitable.
> On the execution side, appropriate operator enhancements would be needed to
> handle different categories of indexes such as covering, non-covering
> indexes, taking into consideration the index data may not be co-located with
> the primary table, i.e a global index.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
