forked from apache/doris
-
Notifications
You must be signed in to change notification settings - Fork 1
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
[feature](Nereids) support materialized index selection (apache#13416)
This PR unified the selection of rollup index and materialized view index into uniform logic, which is called selecting materialized index. Main steps: ### Find candidate indexes 1. When aggregate is present, it's handled in `SelectMaterializedIndexWithAggregate`. The base index and indexes that could use pre-aggregation should be used. The pre-aggregation status is determined by aggregation function, grouping expression, and pushdown predicates. 2. When aggregate is not on top of scan node, it's handled in `SelectMaterializedIndexWithoutAggregate`. The base index and indexes that have all the key columns could be used. ### Filter and order the candidate indexes 1. filter indexes that contain all the required output scan columns. 2. filter indexes that could match prefix index most. 3. order the result index by row count, column count, and index id.
- Loading branch information
1 parent
bd884d3
commit d6c3470
Showing
13 changed files
with
1,798 additions
and
410 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
273 changes: 273 additions & 0 deletions
273
.../src/main/java/org/apache/doris/nereids/rules/mv/AbstractSelectMaterializedIndexRule.java
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,273 @@ | ||
// 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.doris.nereids.rules.mv; | ||
|
||
import org.apache.doris.catalog.Column; | ||
import org.apache.doris.catalog.MaterializedIndex; | ||
import org.apache.doris.catalog.OlapTable; | ||
import org.apache.doris.nereids.trees.expressions.ComparisonPredicate; | ||
import org.apache.doris.nereids.trees.expressions.EqualTo; | ||
import org.apache.doris.nereids.trees.expressions.ExprId; | ||
import org.apache.doris.nereids.trees.expressions.Expression; | ||
import org.apache.doris.nereids.trees.expressions.InPredicate; | ||
import org.apache.doris.nereids.trees.expressions.NamedExpression; | ||
import org.apache.doris.nereids.trees.expressions.NullSafeEqual; | ||
import org.apache.doris.nereids.trees.expressions.Slot; | ||
import org.apache.doris.nereids.trees.expressions.literal.Literal; | ||
import org.apache.doris.nereids.trees.expressions.visitor.ExpressionVisitor; | ||
import org.apache.doris.nereids.trees.plans.logical.LogicalOlapScan; | ||
import org.apache.doris.nereids.util.ExpressionUtils; | ||
|
||
import com.google.common.collect.ImmutableSet; | ||
import com.google.common.collect.Lists; | ||
|
||
import java.util.Comparator; | ||
import java.util.List; | ||
import java.util.Map; | ||
import java.util.Optional; | ||
import java.util.Set; | ||
import java.util.TreeMap; | ||
import java.util.function.Function; | ||
import java.util.stream.Collectors; | ||
import java.util.stream.Stream; | ||
|
||
/** | ||
* Base class for selecting materialized index rules. | ||
*/ | ||
public abstract class AbstractSelectMaterializedIndexRule { | ||
protected boolean shouldSelectIndex(LogicalOlapScan scan) { | ||
switch (scan.getTable().getKeysType()) { | ||
case AGG_KEYS: | ||
case UNIQUE_KEYS: | ||
case DUP_KEYS: | ||
return !scan.isIndexSelected(); | ||
default: | ||
return false; | ||
} | ||
} | ||
|
||
/** | ||
* 1. indexes have all the required columns. | ||
* 2. find matching key prefix most. | ||
* 3. sort by row count, column count and index id. | ||
*/ | ||
protected List<Long> filterAndOrder( | ||
Stream<MaterializedIndex> candidates, | ||
LogicalOlapScan scan, | ||
Set<Slot> requiredScanOutput, | ||
List<Expression> predicates) { | ||
|
||
OlapTable table = scan.getTable(); | ||
// Scan slot exprId -> slot name | ||
Map<ExprId, String> exprIdToName = scan.getOutput() | ||
.stream() | ||
.collect(Collectors.toMap(NamedExpression::getExprId, NamedExpression::getName)); | ||
|
||
// get required column names in metadata. | ||
Set<String> requiredColumnNames = requiredScanOutput | ||
.stream() | ||
.map(slot -> exprIdToName.get(slot.getExprId())) | ||
.collect(Collectors.toSet()); | ||
|
||
// 1. filter index contains all the required columns by column name. | ||
List<MaterializedIndex> containAllRequiredColumns = candidates | ||
.filter(index -> table.getSchemaByIndexId(index.getId(), true) | ||
.stream() | ||
.map(Column::getName) | ||
.collect(Collectors.toSet()) | ||
.containsAll(requiredColumnNames) | ||
).collect(Collectors.toList()); | ||
|
||
// 2. find matching key prefix most. | ||
List<MaterializedIndex> matchingKeyPrefixMost = matchPrefixMost(scan, containAllRequiredColumns, predicates, | ||
exprIdToName); | ||
|
||
List<Long> partitionIds = scan.getSelectedPartitionIds(); | ||
// 3. sort by row count, column count and index id. | ||
return matchingKeyPrefixMost.stream() | ||
.map(MaterializedIndex::getId) | ||
.sorted(Comparator | ||
// compare by row count | ||
.comparing(rid -> partitionIds.stream() | ||
.mapToLong(pid -> table.getPartition(pid).getIndex((Long) rid).getRowCount()) | ||
.sum()) | ||
// compare by column count | ||
.thenComparing(rid -> table.getSchemaByIndexId((Long) rid).size()) | ||
// compare by index id | ||
.thenComparing(rid -> (Long) rid)) | ||
.collect(Collectors.toList()); | ||
} | ||
|
||
protected List<MaterializedIndex> matchPrefixMost( | ||
LogicalOlapScan scan, | ||
List<MaterializedIndex> candidate, | ||
List<Expression> predicates, | ||
Map<ExprId, String> exprIdToName) { | ||
Map<Boolean, Set<String>> split = filterCanUsePrefixIndexAndSplitByEquality(predicates, exprIdToName); | ||
Set<String> equalColNames = split.getOrDefault(true, ImmutableSet.of()); | ||
Set<String> nonEqualColNames = split.getOrDefault(false, ImmutableSet.of()); | ||
|
||
if (!(equalColNames.isEmpty() && nonEqualColNames.isEmpty())) { | ||
List<MaterializedIndex> matchingResult = matchKeyPrefixMost(scan.getTable(), candidate, | ||
equalColNames, nonEqualColNames); | ||
return matchingResult.isEmpty() ? candidate : matchingResult; | ||
} else { | ||
return candidate; | ||
} | ||
} | ||
|
||
/////////////////////////////////////////////////////////////////////////// | ||
// Split conjuncts into equal-to and non-equal-to. | ||
/////////////////////////////////////////////////////////////////////////// | ||
|
||
/** | ||
* Filter the input conjuncts those can use prefix and split into 2 groups: is equal-to or non-equal-to predicate | ||
* when comparing the key column. | ||
*/ | ||
private Map<Boolean, Set<String>> filterCanUsePrefixIndexAndSplitByEquality( | ||
List<Expression> conjunct, Map<ExprId, String> exprIdToColName) { | ||
return conjunct.stream() | ||
.map(expr -> PredicateChecker.canUsePrefixIndex(expr, exprIdToColName)) | ||
.filter(result -> !result.equals(PrefixIndexCheckResult.FAILURE)) | ||
.collect(Collectors.groupingBy( | ||
result -> result.type == ResultType.SUCCESS_EQUAL, | ||
Collectors.mapping(result -> result.colName, Collectors.toSet()) | ||
)); | ||
} | ||
|
||
private enum ResultType { | ||
FAILURE, | ||
SUCCESS_EQUAL, | ||
SUCCESS_NON_EQUAL, | ||
} | ||
|
||
private static class PrefixIndexCheckResult { | ||
public static final PrefixIndexCheckResult FAILURE = new PrefixIndexCheckResult(null, ResultType.FAILURE); | ||
private final String colName; | ||
private final ResultType type; | ||
|
||
private PrefixIndexCheckResult(String colName, ResultType result) { | ||
this.colName = colName; | ||
this.type = result; | ||
} | ||
|
||
public static PrefixIndexCheckResult createEqual(String name) { | ||
return new PrefixIndexCheckResult(name, ResultType.SUCCESS_EQUAL); | ||
} | ||
|
||
public static PrefixIndexCheckResult createNonEqual(String name) { | ||
return new PrefixIndexCheckResult(name, ResultType.SUCCESS_NON_EQUAL); | ||
} | ||
} | ||
|
||
/** | ||
* Check if an expression could prefix key index. | ||
*/ | ||
private static class PredicateChecker extends ExpressionVisitor<PrefixIndexCheckResult, Map<ExprId, String>> { | ||
private static final PredicateChecker INSTANCE = new PredicateChecker(); | ||
|
||
private PredicateChecker() { | ||
} | ||
|
||
public static PrefixIndexCheckResult canUsePrefixIndex(Expression expression, | ||
Map<ExprId, String> exprIdToName) { | ||
return expression.accept(INSTANCE, exprIdToName); | ||
} | ||
|
||
@Override | ||
public PrefixIndexCheckResult visit(Expression expr, Map<ExprId, String> context) { | ||
return PrefixIndexCheckResult.FAILURE; | ||
} | ||
|
||
@Override | ||
public PrefixIndexCheckResult visitInPredicate(InPredicate in, Map<ExprId, String> context) { | ||
Optional<ExprId> slotOrCastOnSlot = ExpressionUtils.isSlotOrCastOnSlot(in.getCompareExpr()); | ||
if (slotOrCastOnSlot.isPresent() && in.getOptions().stream().allMatch(Literal.class::isInstance)) { | ||
return PrefixIndexCheckResult.createEqual(context.get(slotOrCastOnSlot.get())); | ||
} else { | ||
return PrefixIndexCheckResult.FAILURE; | ||
} | ||
} | ||
|
||
@Override | ||
public PrefixIndexCheckResult visitComparisonPredicate(ComparisonPredicate cp, Map<ExprId, String> context) { | ||
if (cp instanceof EqualTo || cp instanceof NullSafeEqual) { | ||
return check(cp, context, PrefixIndexCheckResult::createEqual); | ||
} else { | ||
return check(cp, context, PrefixIndexCheckResult::createNonEqual); | ||
} | ||
} | ||
|
||
private PrefixIndexCheckResult check(ComparisonPredicate cp, Map<ExprId, String> exprIdToColumnName, | ||
Function<String, PrefixIndexCheckResult> resultMapper) { | ||
return check(cp).map(exprId -> resultMapper.apply(exprIdToColumnName.get(exprId))) | ||
.orElse(PrefixIndexCheckResult.FAILURE); | ||
} | ||
|
||
private Optional<ExprId> check(ComparisonPredicate cp) { | ||
Optional<ExprId> exprId = check(cp.left(), cp.right()); | ||
return exprId.isPresent() ? exprId : check(cp.right(), cp.left()); | ||
} | ||
|
||
private Optional<ExprId> check(Expression maybeSlot, Expression maybeConst) { | ||
Optional<ExprId> exprIdOpt = ExpressionUtils.isSlotOrCastOnSlot(maybeSlot); | ||
return exprIdOpt.isPresent() && maybeConst.isConstant() ? exprIdOpt : Optional.empty(); | ||
} | ||
} | ||
|
||
/////////////////////////////////////////////////////////////////////////// | ||
// Matching key prefix | ||
/////////////////////////////////////////////////////////////////////////// | ||
private List<MaterializedIndex> matchKeyPrefixMost( | ||
OlapTable table, | ||
List<MaterializedIndex> indexes, | ||
Set<String> equalColumns, | ||
Set<String> nonEqualColumns) { | ||
TreeMap<Integer, List<MaterializedIndex>> collect = indexes.stream() | ||
.collect(Collectors.toMap( | ||
index -> indexKeyPrefixMatchCount(table, index, equalColumns, nonEqualColumns), | ||
Lists::newArrayList, | ||
(l1, l2) -> { | ||
l1.addAll(l2); | ||
return l1; | ||
}, | ||
TreeMap::new) | ||
); | ||
return collect.descendingMap().firstEntry().getValue(); | ||
} | ||
|
||
private int indexKeyPrefixMatchCount( | ||
OlapTable table, | ||
MaterializedIndex index, | ||
Set<String> equalColNames, | ||
Set<String> nonEqualColNames) { | ||
int matchCount = 0; | ||
for (Column column : table.getSchemaByIndexId(index.getId())) { | ||
if (equalColNames.contains(column.getName())) { | ||
matchCount++; | ||
} else if (nonEqualColNames.contains(column.getName())) { | ||
// Unequivalence predicate's columns can match only first column in index. | ||
matchCount++; | ||
break; | ||
} else { | ||
break; | ||
} | ||
} | ||
return matchCount; | ||
} | ||
} |
Oops, something went wrong.