Skip to content

Commit

Permalink
Enable Fuzzy codec for doc id fields using a bloom filter
Browse files Browse the repository at this point in the history
Signed-off-by: mgodwan <mgodwan@amazon.com>
  • Loading branch information
mgodwan committed Oct 31, 2023
1 parent a2febe9 commit a10dd7f
Show file tree
Hide file tree
Showing 19 changed files with 1,387 additions and 1 deletion.
Original file line number Diff line number Diff line change
@@ -0,0 +1,67 @@
/*
* SPDX-License-Identifier: Apache-2.0
*
* The OpenSearch Contributors require contributions made to
* this file be licensed under the Apache-2.0 license or a
* compatible open source license.
*/

package org.opensearch.benchmark.index.codec.fuzzy;

import org.apache.lucene.util.BytesRef;
import org.opensearch.common.UUIDs;
import org.opensearch.index.codec.fuzzy.FuzzySet;
import org.opensearch.index.codec.fuzzy.FuzzySetFactory;
import org.opensearch.index.codec.fuzzy.FuzzySetParameters;
import org.opensearch.index.mapper.IdFieldMapper;
import org.openjdk.jmh.annotations.Benchmark;
import org.openjdk.jmh.annotations.BenchmarkMode;
import org.openjdk.jmh.annotations.Fork;
import org.openjdk.jmh.annotations.Measurement;
import org.openjdk.jmh.annotations.Mode;
import org.openjdk.jmh.annotations.OutputTimeUnit;
import org.openjdk.jmh.annotations.Param;
import org.openjdk.jmh.annotations.Scope;
import org.openjdk.jmh.annotations.Setup;
import org.openjdk.jmh.annotations.State;
import org.openjdk.jmh.annotations.Warmup;

import java.io.IOException;
import java.util.List;
import java.util.Map;
import java.util.concurrent.TimeUnit;
import java.util.stream.Collectors;
import java.util.stream.IntStream;

@Fork(3)
@Warmup(iterations = 2)
@Measurement(iterations = 5, time = 60, timeUnit = TimeUnit.SECONDS)
@BenchmarkMode(Mode.AverageTime)
@OutputTimeUnit(TimeUnit.MILLISECONDS)
@State(Scope.Benchmark)
public class FilterConstructionBenchmark {

private List<BytesRef> items;

@Param({ "1000000", "10000000", "50000000" })
private int numIds;

@Param({ "0.0511", "0.1023", "0.2047" })
private double fpp;

private FuzzySetFactory fuzzySetFactory;
private String fieldName;

@Setup
public void setupIds() {
this.fieldName = IdFieldMapper.NAME;
this.items = IntStream.range(0, numIds).mapToObj(i -> new BytesRef(UUIDs.base64UUID())).collect(Collectors.toList());
FuzzySetParameters parameters = new FuzzySetParameters(() -> fpp);
this.fuzzySetFactory = new FuzzySetFactory(Map.of(fieldName, parameters));
}

@Benchmark
public FuzzySet buildFilter() throws IOException {
return fuzzySetFactory.createFuzzySet(items.size(), fieldName, () -> items.iterator());
}
}
Original file line number Diff line number Diff line change
@@ -0,0 +1,80 @@
/*
* SPDX-License-Identifier: Apache-2.0
*
* The OpenSearch Contributors require contributions made to
* this file be licensed under the Apache-2.0 license or a
* compatible open source license.
*/

package org.opensearch.benchmark.index.codec.fuzzy;

import org.apache.lucene.util.BytesRef;
import org.opensearch.common.UUIDs;
import org.opensearch.index.codec.fuzzy.FuzzySet;
import org.opensearch.index.codec.fuzzy.FuzzySetFactory;
import org.opensearch.index.codec.fuzzy.FuzzySetParameters;
import org.opensearch.index.mapper.IdFieldMapper;
import org.openjdk.jmh.annotations.Benchmark;
import org.openjdk.jmh.annotations.BenchmarkMode;
import org.openjdk.jmh.annotations.Fork;
import org.openjdk.jmh.annotations.Measurement;
import org.openjdk.jmh.annotations.Mode;
import org.openjdk.jmh.annotations.OutputTimeUnit;
import org.openjdk.jmh.annotations.Param;
import org.openjdk.jmh.annotations.Scope;
import org.openjdk.jmh.annotations.Setup;
import org.openjdk.jmh.annotations.State;
import org.openjdk.jmh.annotations.Warmup;
import org.openjdk.jmh.infra.Blackhole;

import java.io.IOException;
import java.util.List;
import java.util.Map;
import java.util.Random;
import java.util.concurrent.TimeUnit;
import java.util.stream.Collectors;
import java.util.stream.IntStream;

@Fork(3)
@Warmup(iterations = 2)
@Measurement(iterations = 5, time = 60, timeUnit = TimeUnit.SECONDS)
@BenchmarkMode(Mode.AverageTime)
@OutputTimeUnit(TimeUnit.MILLISECONDS)
@State(Scope.Benchmark)
public class FilterLookupBenchmark {

@Param({ "50000000", "1000000" })
private int numItems;

@Param({ "1000000" })
private int searchKeyCount;

@Param({ "0.0511", "0.1023", "0.2047" })
private double fpp;

private FuzzySet fuzzySet;
private List<BytesRef> items;
private Random random = new Random();

@Setup
public void setupFilter() throws IOException {
String fieldName = IdFieldMapper.NAME;
items = IntStream.range(0, numItems).mapToObj(i -> new BytesRef(UUIDs.base64UUID())).collect(Collectors.toList());
FuzzySetParameters parameters = new FuzzySetParameters(() -> fpp);
fuzzySet = new FuzzySetFactory(Map.of(fieldName, parameters)).createFuzzySet(numItems, fieldName, () -> items.iterator());
}

@Benchmark
public void contains_withExistingKeys(Blackhole blackhole) throws IOException {
for (int i = 0; i < searchKeyCount; i++) {
blackhole.consume(fuzzySet.contains(items.get(random.nextInt(items.size()))) == FuzzySet.Result.MAYBE);
}
}

@Benchmark
public void contains_withRandomKeys(Blackhole blackhole) throws IOException {
for (int i = 0; i < searchKeyCount; i++) {
blackhole.consume(fuzzySet.contains(new BytesRef(UUIDs.base64UUID())));
}
}
}
Original file line number Diff line number Diff line change
Expand Up @@ -228,6 +228,9 @@ public final class IndexScopedSettings extends AbstractScopedSettings {
IndexMetadata.INDEX_REMOTE_SEGMENT_STORE_REPOSITORY_SETTING,
IndexMetadata.INDEX_REMOTE_TRANSLOG_REPOSITORY_SETTING,

IndexSettings.INDEX_DOC_ID_FUZZY_SET_ENABLED_SETTING,
IndexSettings.INDEX_DOC_ID_FUZZY_SET_FALSE_POSITIVE_PROBABILITY_SETTING,

// validate that built-in similarities don't get redefined
Setting.groupSetting("index.similarity.", (s) -> {
Map<String, Settings> groups = s.getAsGroups();
Expand Down
50 changes: 50 additions & 0 deletions server/src/main/java/org/opensearch/index/IndexSettings.java
Original file line number Diff line number Diff line change
Expand Up @@ -66,6 +66,7 @@

import static org.opensearch.Version.V_2_7_0;
import static org.opensearch.common.util.FeatureFlags.SEARCHABLE_SNAPSHOT_EXTENDED_COMPATIBILITY;
import static org.opensearch.index.codec.fuzzy.FuzzySetParameters.DEFAULT_FALSE_POSITIVE_PROBABILITY;
import static org.opensearch.index.mapper.MapperService.INDEX_MAPPING_DEPTH_LIMIT_SETTING;
import static org.opensearch.index.mapper.MapperService.INDEX_MAPPING_FIELD_NAME_LENGTH_LIMIT_SETTING;
import static org.opensearch.index.mapper.MapperService.INDEX_MAPPING_NESTED_DOCS_LIMIT_SETTING;
Expand Down Expand Up @@ -658,6 +659,22 @@ public static IndexMergePolicy fromString(String text) {
Property.Dynamic
);

public static final Setting<Boolean> INDEX_DOC_ID_FUZZY_SET_ENABLED_SETTING = Setting.boolSetting(
"index.doc_id_fuzzy_set.enabled",
false,
Property.IndexScope,
Property.Dynamic
);

public static final Setting<Double> INDEX_DOC_ID_FUZZY_SET_FALSE_POSITIVE_PROBABILITY_SETTING = Setting.doubleSetting(
"index.doc_id_fuzzy_set.false_positive_probability",
DEFAULT_FALSE_POSITIVE_PROBABILITY,
0.01,
0.50,
Property.IndexScope,
Property.Dynamic
);

public static final TimeValue DEFAULT_REMOTE_TRANSLOG_BUFFER_INTERVAL = new TimeValue(650, TimeUnit.MILLISECONDS);
public static final TimeValue MINIMUM_REMOTE_TRANSLOG_BUFFER_INTERVAL = TimeValue.ZERO;
public static final Setting<TimeValue> INDEX_REMOTE_TRANSLOG_BUFFER_INTERVAL_SETTING = Setting.timeSetting(
Expand Down Expand Up @@ -787,6 +804,17 @@ private void setRetentionLeaseMillis(final TimeValue retentionLease) {
*/
private volatile UnaryOperator<MergePolicy> mergeOnFlushPolicy;

/**
* Is fuzzy set enabled for doc id
*/
private volatile boolean enableFuzzySetForDocId;

/**
* False positive probability to use while creating fuzzy set.
*/
private volatile double docIdFuzzySetFalsePositiveProbability;


/**
* Returns the default search fields for this index.
*/
Expand Down Expand Up @@ -926,6 +954,10 @@ public IndexSettings(final IndexMetadata indexMetadata, final Settings nodeSetti
* Now this sortField (IndexSort) is stored in SegmentInfo and we need to maintain backward compatibility for them.
*/
widenIndexSortType = IndexMetadata.SETTING_INDEX_VERSION_CREATED.get(settings).before(V_2_7_0);

enableFuzzySetForDocId = scopedSettings.get(INDEX_DOC_ID_FUZZY_SET_ENABLED_SETTING);
docIdFuzzySetFalsePositiveProbability = scopedSettings.get(INDEX_DOC_ID_FUZZY_SET_FALSE_POSITIVE_PROBABILITY_SETTING);

scopedSettings.addSettingsUpdateConsumer(
TieredMergePolicyProvider.INDEX_COMPOUND_FORMAT_SETTING,
tieredMergePolicyProvider::setNoCFSRatio
Expand Down Expand Up @@ -1032,6 +1064,8 @@ public IndexSettings(final IndexMetadata indexMetadata, final Settings nodeSetti
this::setRemoteTranslogUploadBufferInterval
);
scopedSettings.addSettingsUpdateConsumer(INDEX_REMOTE_TRANSLOG_KEEP_EXTRA_GEN_SETTING, this::setRemoteTranslogKeepExtraGen);
scopedSettings.addSettingsUpdateConsumer(INDEX_DOC_ID_FUZZY_SET_ENABLED_SETTING, this::setEnableFuzzySetForDocId);
scopedSettings.addSettingsUpdateConsumer(INDEX_DOC_ID_FUZZY_SET_FALSE_POSITIVE_PROBABILITY_SETTING, this::setDocIdFuzzySetFalsePositiveProbability);
}

private void setSearchIdleAfter(TimeValue searchIdleAfter) {
Expand Down Expand Up @@ -1801,4 +1835,20 @@ public void setDefaultSearchPipeline(String defaultSearchPipeline) {
public boolean shouldWidenIndexSortType() {
return this.widenIndexSortType;
}

public boolean isEnableFuzzySetForDocId() {
return enableFuzzySetForDocId;
}

public void setEnableFuzzySetForDocId(boolean enableFuzzySetForDocId) {
this.enableFuzzySetForDocId = enableFuzzySetForDocId;
}

public double getDocIdFuzzySetFalsePositiveProbability() {
return docIdFuzzySetFalsePositiveProbability;
}

public void setDocIdFuzzySetFalsePositiveProbability(double docIdFuzzySetFalsePositiveProbability) {
this.docIdFuzzySetFalsePositiveProbability = docIdFuzzySetFalsePositiveProbability;
}
}
Original file line number Diff line number Diff line change
Expand Up @@ -39,10 +39,16 @@
import org.apache.lucene.codecs.lucene90.Lucene90DocValuesFormat;
import org.apache.lucene.codecs.lucene95.Lucene95Codec;
import org.opensearch.common.lucene.Lucene;
import org.opensearch.index.codec.fuzzy.FuzzyFilterPostingsFormat;
import org.opensearch.index.codec.fuzzy.FuzzySetFactory;
import org.opensearch.index.codec.fuzzy.FuzzySetParameters;
import org.opensearch.index.mapper.CompletionFieldMapper;
import org.opensearch.index.mapper.IdFieldMapper;
import org.opensearch.index.mapper.MappedFieldType;
import org.opensearch.index.mapper.MapperService;

import java.util.Map;

/**
* {@link PerFieldMappingPostingFormatCodec This postings format} is the default
* {@link PostingsFormat} for OpenSearch. It utilizes the
Expand All @@ -57,6 +63,8 @@ public class PerFieldMappingPostingFormatCodec extends Lucene95Codec {
private final Logger logger;
private final MapperService mapperService;
private final DocValuesFormat dvFormat = new Lucene90DocValuesFormat();
private FuzzySetFactory fuzzySetFactory;
private FuzzyFilterPostingsFormat docIdPostingsFormat;

static {
assert Codec.forName(Lucene.LATEST_CODEC).getClass().isAssignableFrom(PerFieldMappingPostingFormatCodec.class)
Expand All @@ -67,6 +75,8 @@ public PerFieldMappingPostingFormatCodec(Mode compressionMode, MapperService map
super(compressionMode);
this.mapperService = mapperService;
this.logger = logger;
fuzzySetFactory = new FuzzySetFactory(Map.of(IdFieldMapper.NAME,
new FuzzySetParameters(() -> mapperService.getIndexSettings().getDocIdFuzzySetFalsePositiveProbability())));
}

@Override
Expand All @@ -76,6 +86,11 @@ public PostingsFormat getPostingsFormatForField(String field) {
logger.warn("no index mapper found for field: [{}] returning default postings format", field);
} else if (fieldType instanceof CompletionFieldMapper.CompletionFieldType) {
return CompletionFieldMapper.CompletionFieldType.postingsFormat();
} else if (IdFieldMapper.NAME.equals(field) && mapperService.getIndexSettings().isEnableFuzzySetForDocId()) {
if (docIdPostingsFormat == null) {
docIdPostingsFormat = new FuzzyFilterPostingsFormat(super.getPostingsFormatForField(field), fuzzySetFactory);
}
return docIdPostingsFormat;
}
return super.getPostingsFormatForField(field);
}
Expand Down
Original file line number Diff line number Diff line change
@@ -0,0 +1,52 @@
/*
* SPDX-License-Identifier: Apache-2.0
*
* The OpenSearch Contributors require contributions made to
* this file be licensed under the Apache-2.0 license or a
* compatible open source license.
*/

package org.opensearch.index.codec.fuzzy;

import org.apache.lucene.util.BytesRef;
import org.opensearch.common.CheckedSupplier;

import java.io.IOException;
import java.util.ArrayList;
import java.util.Iterator;
import java.util.List;

public abstract class AbstractFuzzySet implements FuzzySet {

/**
* Add an item to this fuzzy set.
* @param value The value to be added
*/
protected abstract void add(BytesRef value);

/**
* Add all items to the underlying set.
* Implementations can choose to perform this using an optimized strategy based on the type of set.
* @param valuesIteratorProvider Supplier for an iterator over All values which should be added to the set.
*/
protected void addAll(CheckedSupplier<Iterator<BytesRef>, IOException> valuesIteratorProvider) throws IOException {
Iterator<BytesRef> values = valuesIteratorProvider.get();
while (values.hasNext()) {
add(values.next());
}
}

protected long generateKey(BytesRef value) {
return MurmurHash64.INSTANCE.hash(value);
}

protected void assertAllElementsExist(CheckedSupplier<Iterator<BytesRef>, IOException> iteratorProvider) throws IOException {
Iterator<BytesRef> iter = iteratorProvider.get();
int cnt = 0;
while (iter.hasNext()) {
BytesRef item = iter.next();
assert contains(item) == Result.MAYBE : "Expected Filter to return positive response for elements added to it. Elements matched: " + cnt;
cnt ++;
}
}
}

0 comments on commit a10dd7f

Please sign in to comment.