ahmedabu98 commented on code in PR #33504: URL: https://github.com/apache/beam/pull/33504#discussion_r1987477117
########## sdks/java/io/iceberg/src/main/java/org/apache/beam/sdk/io/iceberg/IncrementalScanSource.java: ########## @@ -0,0 +1,163 @@ +/* + * 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.beam.sdk.io.iceberg; + +import static org.apache.beam.sdk.util.Preconditions.checkStateNotNull; + +import java.util.List; +import org.apache.beam.sdk.coders.IterableCoder; +import org.apache.beam.sdk.coders.KvCoder; +import org.apache.beam.sdk.coders.ListCoder; +import org.apache.beam.sdk.coders.StringUtf8Coder; +import org.apache.beam.sdk.transforms.Create; +import org.apache.beam.sdk.transforms.GroupIntoBatches; +import org.apache.beam.sdk.transforms.MapElements; +import org.apache.beam.sdk.transforms.PTransform; +import org.apache.beam.sdk.transforms.ParDo; +import org.apache.beam.sdk.transforms.Redistribute; +import org.apache.beam.sdk.transforms.SimpleFunction; +import org.apache.beam.sdk.transforms.windowing.AfterPane; +import org.apache.beam.sdk.transforms.windowing.GlobalWindows; +import org.apache.beam.sdk.transforms.windowing.Repeatedly; +import org.apache.beam.sdk.transforms.windowing.Window; +import org.apache.beam.sdk.util.ShardedKey; +import org.apache.beam.sdk.values.KV; +import org.apache.beam.sdk.values.PBegin; +import org.apache.beam.sdk.values.PCollection; +import org.apache.beam.sdk.values.Row; +import org.apache.beam.vendor.guava.v32_1_2_jre.com.google.common.base.MoreObjects; +import org.apache.beam.vendor.guava.v32_1_2_jre.com.google.common.collect.Lists; +import org.apache.iceberg.Table; +import org.checkerframework.checker.nullness.qual.Nullable; +import org.joda.time.Duration; + +/** + * An Iceberg source that reads a table incrementally using range(s) of table snapshots. The bounded + * source creates a single range, while the unbounded implementation continuously polls for new + * snapshots at the specified interval. + * + * <p>Outputs CDC Rows with the following schema: + * + * <ul> + * <li>"record" [Row]: the data record itself + * <li>"operation" [String]: the operation associated with this record + * </ul> + */ +class IncrementalScanSource extends PTransform<PBegin, PCollection<Row>> { + // (streaming) We will group files into batches of this size + // Downstream, we create one reader per batch + // TODO(ahmedabu98): should we make this configurable? + private static final long MAX_FILES_BATCH_BYTE_SIZE = 1L << 32; // 4 GB + private static final Duration DEFAULT_POLL_INTERVAL = Duration.standardSeconds(60); + private final IcebergScanConfig scanConfig; + + IncrementalScanSource(IcebergScanConfig scanConfig) { + this.scanConfig = scanConfig; + } + + @Override + public PCollection<Row> expand(PBegin input) { + Table table = + TableCache.getRefreshed( + scanConfig.getTableIdentifier(), scanConfig.getCatalogConfig().catalog()); + + PCollection<Row> rows = + MoreObjects.firstNonNull(scanConfig.getStreaming(), false) + ? readUnbounded(input) + : readBounded(input, table); + return rows.setRowSchema(ReadUtils.outputCdcSchema(table.schema())); + } + + /** + * Watches for new snapshots and creates tasks for each range. Uses GiB (with auto-sharding) to + * groups tasks in batches of size {@link #MAX_FILES_BATCH_BYTE_SIZE}, then reads from each batch + * using an SDF. + * + * <p>Output schema is: + * + * <ul> + * <li>"record": the actual data record + * <li>"operation": the snapshot operation associated with this record (e.g. "append", + * "replace", "delete") + * </ul> + */ + private PCollection<Row> readUnbounded(PBegin input) { + @Nullable + Duration pollInterval = + MoreObjects.firstNonNull(scanConfig.getPollInterval(), DEFAULT_POLL_INTERVAL); + return input + .apply("Watch for Snapshots", new WatchForSnapshots(scanConfig, pollInterval)) + .setCoder(KvCoder.of(StringUtf8Coder.of(), ListCoder.of(SnapshotInfo.getCoder()))) + .apply("Create Read Tasks", ParDo.of(new CreateReadTasksDoFn(scanConfig))) + .setCoder(KvCoder.of(ReadTaskDescriptor.getCoder(), ReadTask.getCoder())) + .apply( + Window.<KV<ReadTaskDescriptor, ReadTask>>into(new GlobalWindows()) + .triggering(Repeatedly.forever(AfterPane.elementCountAtLeast(1))) + .discardingFiredPanes()) + .apply( + GroupIntoBatches.<ReadTaskDescriptor, ReadTask>ofByteSize( + MAX_FILES_BATCH_BYTE_SIZE, ReadTask::getByteSize) Review Comment: I also tried mimicking GiB behavior by: - Associating a key to each read task. The key is incremented after reaching 4GB. - Adding a Redistribute.byKey() after CreateReadTasksDoFn to distribute read tasks into 4GB per streaming key But the throughput was pretty spiky and took much longer than the current approach: [2025-03-07_08_57_21-15760437490773458424](https://dataflow-console.corp.google.com/?justification=#/jobmanagement/2025-03-07_08_57_21-15760437490773458424) -- 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: github-unsubscr...@beam.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
