zhongyujiang commented on code in PR #9177: URL: https://github.com/apache/iceberg/pull/9177#discussion_r1408832480
########## spark/v3.5/spark/src/jmh/java/org/apache/iceberg/spark/source/IcebergSourceParquetPosDeleteRowGroupFilterBenchmark.java: ########## @@ -0,0 +1,125 @@ +/* + * 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.iceberg.spark.source; + +import java.io.IOException; +import java.util.List; +import java.util.stream.Collectors; +import org.apache.iceberg.DeleteFile; +import org.apache.iceberg.FileFormat; +import org.apache.iceberg.Files; +import org.apache.iceberg.MetadataColumns; +import org.apache.iceberg.Schema; +import org.apache.iceberg.TableProperties; +import org.apache.iceberg.data.Record; +import org.apache.iceberg.data.parquet.GenericParquetReaders; +import org.apache.iceberg.io.CloseableIterable; +import org.apache.iceberg.parquet.Parquet; +import org.apache.iceberg.relocated.com.google.common.collect.Iterables; +import org.apache.iceberg.relocated.com.google.common.collect.Streams; +import org.openjdk.jmh.annotations.Benchmark; +import org.openjdk.jmh.annotations.Param; +import org.openjdk.jmh.annotations.Threads; +import org.openjdk.jmh.infra.Blackhole; +import org.slf4j.Logger; +import org.slf4j.LoggerFactory; + +/** + * A benchmark that evaluates the non-vectorized read and vectorized read with pos-delete in the + * Spark data source for Iceberg. + * + * <p>This class uses a dataset with a flat schema. To run this benchmark for spark-3.5: <code> + * ./gradlew -DsparkVersions=3.5 :iceberg-spark:iceberg-spark-3.5:jmh + * -PjmhIncludeRegex=IcebergSourceParquetPosDeleteRowGroupFilterBenchmark.readIcebergVectorized + * -PjmhOutputPath=benchmark/iceberg-source-parquet-pos-delete-row-group-filter-benchmark-result.txt + * </code> + */ +public class IcebergSourceParquetPosDeleteRowGroupFilterBenchmark + extends IcebergSourceDeleteBenchmark { + private static final Logger LOG = + LoggerFactory.getLogger(IcebergSourceParquetPosDeleteRowGroupFilterBenchmark.class); + + private static final Schema DELETE_SCHEMA = + new Schema(MetadataColumns.DELETE_FILE_PATH, MetadataColumns.DELETE_FILE_POS); + + @Param({"2", "4", "128" /* default size */}) + private int rowGroupSizeMB; + + @Param({"3", "5"}) + private int numDataFiles; + + private DeleteFile posDeleteFile; + + @Override + protected void appendData() throws IOException { + for (int fileNum = 1; fileNum <= numDataFiles; fileNum++) { + writeData(fileNum); + } + + table() + .updateProperties() + .set( + TableProperties.PATH_POS_DELETE_PARQUET_ROW_GROUP_SIZE_BYTES, + String.valueOf(rowGroupSizeMB * 1024 * 1024)) + .commit(); + + // add pos-deletes + table().refresh(); + List<CharSequence> filePaths = + Streams.stream(table().newScan().planFiles()) + .map(task -> task.file().path()) + .collect(Collectors.toList()); + + writePosDeletes(filePaths, NUM_ROWS, 0.10); + + table().refresh(); + posDeleteFile = + Iterables.getOnlyElement(table().currentSnapshot().addedDeleteFiles(table().io())); + List<Long> offsets = posDeleteFile.splitOffsets(); + + LOG.info( + "row-group-size-MB: {}, position delete file size in bytes: {}, number row-groups: {}", + rowGroupSizeMB, + posDeleteFile.fileSizeInBytes(), + offsets.size()); Review Comment: This is the row-group split status of position delete files when using different row-group sizes, the actual row-group size might be smaller than the set value because Parquet will make judgments based on the unflushed data size when splitting row-groups: ``` row-group-size-MB: 128, position delete file size in bytes: 9921119, number row-groups: 1 row-group-size-MB: 4, position delete file size in bytes: 9921030, number row-groups: 3 row-group-size-MB: 2, position delete file size in bytes: 9925059, number row-groups: 8 row-group-size-MB: 128, position delete file size in bytes: 5951966, number row-groups: 1 row-group-size-MB: 4, position delete file size in bytes: 5952190, number row-groups: 2 row-group-size-MB: 2, position delete file size in bytes: 5955593, number row-groups: 5 ``` And below is the benchmark result: ``` Benchmark (numDataFiles) (rowGroupSizeMB) Mode Cnt Score Error Units IcebergSourceParquetPosDeleteRowGroupFilterBenchmark.readIcebergVectorized 3 2 ss 5 4.853 ± 0.632 s/op IcebergSourceParquetPosDeleteRowGroupFilterBenchmark.readIcebergVectorized 3 4 ss 5 5.794 ± 0.379 s/op IcebergSourceParquetPosDeleteRowGroupFilterBenchmark.readIcebergVectorized 3 128 ss 5 6.207 ± 0.278 s/op IcebergSourceParquetPosDeleteRowGroupFilterBenchmark.readIcebergVectorized 5 2 ss 5 8.758 ± 0.610 s/op IcebergSourceParquetPosDeleteRowGroupFilterBenchmark.readIcebergVectorized 5 4 ss 5 10.270 ± 0.586 s/op IcebergSourceParquetPosDeleteRowGroupFilterBenchmark.readIcebergVectorized 5 128 ss 5 13.734 ± 0.433 s/op ``` ########## spark/v3.5/spark/src/jmh/java/org/apache/iceberg/spark/source/IcebergSourceParquetPosDeleteRowGroupFilterBenchmark.java: ########## @@ -0,0 +1,125 @@ +/* + * 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.iceberg.spark.source; + +import java.io.IOException; +import java.util.List; +import java.util.stream.Collectors; +import org.apache.iceberg.DeleteFile; +import org.apache.iceberg.FileFormat; +import org.apache.iceberg.Files; +import org.apache.iceberg.MetadataColumns; +import org.apache.iceberg.Schema; +import org.apache.iceberg.TableProperties; +import org.apache.iceberg.data.Record; +import org.apache.iceberg.data.parquet.GenericParquetReaders; +import org.apache.iceberg.io.CloseableIterable; +import org.apache.iceberg.parquet.Parquet; +import org.apache.iceberg.relocated.com.google.common.collect.Iterables; +import org.apache.iceberg.relocated.com.google.common.collect.Streams; +import org.openjdk.jmh.annotations.Benchmark; +import org.openjdk.jmh.annotations.Param; +import org.openjdk.jmh.annotations.Threads; +import org.openjdk.jmh.infra.Blackhole; +import org.slf4j.Logger; +import org.slf4j.LoggerFactory; + +/** + * A benchmark that evaluates the non-vectorized read and vectorized read with pos-delete in the + * Spark data source for Iceberg. + * + * <p>This class uses a dataset with a flat schema. To run this benchmark for spark-3.5: <code> + * ./gradlew -DsparkVersions=3.5 :iceberg-spark:iceberg-spark-3.5:jmh + * -PjmhIncludeRegex=IcebergSourceParquetPosDeleteRowGroupFilterBenchmark.readIcebergVectorized + * -PjmhOutputPath=benchmark/iceberg-source-parquet-pos-delete-row-group-filter-benchmark-result.txt + * </code> + */ +public class IcebergSourceParquetPosDeleteRowGroupFilterBenchmark + extends IcebergSourceDeleteBenchmark { + private static final Logger LOG = + LoggerFactory.getLogger(IcebergSourceParquetPosDeleteRowGroupFilterBenchmark.class); + + private static final Schema DELETE_SCHEMA = + new Schema(MetadataColumns.DELETE_FILE_PATH, MetadataColumns.DELETE_FILE_POS); + + @Param({"2", "4", "128" /* default size */}) + private int rowGroupSizeMB; + + @Param({"3", "5"}) + private int numDataFiles; + + private DeleteFile posDeleteFile; + + @Override + protected void appendData() throws IOException { + for (int fileNum = 1; fileNum <= numDataFiles; fileNum++) { + writeData(fileNum); + } + + table() + .updateProperties() + .set( + TableProperties.PATH_POS_DELETE_PARQUET_ROW_GROUP_SIZE_BYTES, + String.valueOf(rowGroupSizeMB * 1024 * 1024)) + .commit(); + + // add pos-deletes + table().refresh(); + List<CharSequence> filePaths = + Streams.stream(table().newScan().planFiles()) + .map(task -> task.file().path()) + .collect(Collectors.toList()); + + writePosDeletes(filePaths, NUM_ROWS, 0.10); + + table().refresh(); + posDeleteFile = + Iterables.getOnlyElement(table().currentSnapshot().addedDeleteFiles(table().io())); + List<Long> offsets = posDeleteFile.splitOffsets(); + + LOG.info( + "row-group-size-MB: {}, position delete file size in bytes: {}, number row-groups: {}", + rowGroupSizeMB, + posDeleteFile.fileSizeInBytes(), + offsets.size()); Review Comment: This is the row-group split status of position delete files when using different row-group sizes, the actual row-group size might be smaller than the set value because Parquet will make judgments based on the unflushed data size when splitting row-groups: ``` row-group-size-MB: 128, position delete file size in bytes: 9921119, number row-groups: 1 row-group-size-MB: 4, position delete file size in bytes: 9921030, number row-groups: 3 row-group-size-MB: 2, position delete file size in bytes: 9925059, number row-groups: 8 row-group-size-MB: 128, position delete file size in bytes: 5951966, number row-groups: 1 row-group-size-MB: 4, position delete file size in bytes: 5952190, number row-groups: 2 row-group-size-MB: 2, position delete file size in bytes: 5955593, number row-groups: 5 ``` And below is the benchmark result: ``` Benchmark (numDataFiles) (rowGroupSizeMB) Mode Cnt Score Error Units IcebergSourceParquetPosDeleteRowGroupFilterBenchmark.readIcebergVectorized 3 2 ss 5 4.853 ± 0.632 s/op IcebergSourceParquetPosDeleteRowGroupFilterBenchmark.readIcebergVectorized 3 4 ss 5 5.794 ± 0.379 s/op IcebergSourceParquetPosDeleteRowGroupFilterBenchmark.readIcebergVectorized 3 128 ss 5 6.207 ± 0.278 s/op IcebergSourceParquetPosDeleteRowGroupFilterBenchmark.readIcebergVectorized 5 2 ss 5 8.758 ± 0.610 s/op IcebergSourceParquetPosDeleteRowGroupFilterBenchmark.readIcebergVectorized 5 4 ss 5 10.270 ± 0.586 s/op IcebergSourceParquetPosDeleteRowGroupFilterBenchmark.readIcebergVectorized 5 128 ss 5 13.734 ± 0.433 s/op ``` -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: issues-unsubscr...@iceberg.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@iceberg.apache.org For additional commands, e-mail: issues-h...@iceberg.apache.org
