openinx commented on a change in pull request #3164: URL: https://github.com/apache/iceberg/pull/3164#discussion_r714601749
########## File path: core/src/main/java/org/apache/iceberg/io/FanoutWriter.java ########## @@ -0,0 +1,103 @@ +/* + * 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.io; + +import java.io.IOException; +import java.util.Map; +import org.apache.iceberg.PartitionSpec; +import org.apache.iceberg.StructLike; +import org.apache.iceberg.encryption.EncryptedOutputFile; +import org.apache.iceberg.relocated.com.google.common.base.Preconditions; +import org.apache.iceberg.relocated.com.google.common.collect.Maps; +import org.apache.iceberg.util.StructLikeMap; + +/** + * A writer capable of writing to multiple specs and partitions that keeps files for each + * seen spec/partition pair open until this writer is closed. + * <p> + * As opposed to {@link ClusteredWriter}, this writer does not require the incoming records + * to be clustered by partition spec and partition as all files are kept open. As a consequence, + * this writer may potentially consume substantially more memory compared to {@link ClusteredWriter}. + * Use this writer only when clustering by spec/partition is not possible (e.g. streaming). + */ +abstract class FanoutWriter<T, R> implements PartitioningWriter<T, R> { + + private final Map<Integer, Map<StructLike, FileWriter<T, R>>> writers = Maps.newHashMap(); + private boolean closed = false; + + protected abstract FileWriter<T, R> newWriter(PartitionSpec spec, StructLike partition); + + protected abstract void addResult(R result); + + protected abstract R aggregatedResult(); + + @Override + public void write(T row, PartitionSpec spec, StructLike partition) throws IOException { + FileWriter<T, R> writer = writer(spec, partition); + writer.write(row); + } + + private FileWriter<T, R> writer(PartitionSpec spec, StructLike partition) { + Map<StructLike, FileWriter<T, R>> specWriters = writers.computeIfAbsent( + spec.specId(), + id -> StructLikeMap.create(spec.partitionType())); + FileWriter<T, R> writer = specWriters.get(partition); Review comment: For fanout write path, this line is the one of the hottest line because it will need to compare the partition field values for every row. For unpartitioned table, we also need to get the null key from the `specWriters` map. In the old implementation, we don't need to get the writer from the map for unpartitioned table. Is there any performance regression when comparing the two ? -- 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: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
