viirya commented on code in PR #55420: URL: https://github.com/apache/spark/pull/55420#discussion_r3278041941
########## connector/kafka-0-10-sql/src/test/scala/org/apache/spark/sql/kafka010/benchmark/RTMKafkaKafkaBenchmark.scala: ########## @@ -0,0 +1,433 @@ +/* + * 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.spark.sql.kafka010.benchmark + +import java.io.File +import java.util.{Properties, Timer, TimerTask} +import java.util.concurrent.{CountDownLatch, TimeUnit} +import java.util.concurrent.atomic.{AtomicInteger, AtomicLong} + +import scala.concurrent.duration._ + +import org.apache.kafka.clients.producer.{Callback, KafkaProducer, Producer, ProducerRecord, RecordMetadata} + +import org.apache.spark.benchmark.{Benchmark, BenchmarkBase} +import org.apache.spark.internal.Logging +import org.apache.spark.sql.{Column, SparkSession} +import org.apache.spark.sql.execution.streaming.RealTimeTrigger +import org.apache.spark.sql.functions._ +import org.apache.spark.sql.internal.SQLConf +import org.apache.spark.sql.kafka010.KafkaTestUtils +import org.apache.spark.sql.streaming.StreamingQueryListener +import org.apache.spark.util.Utils + +/** + * Stateless Kafka-to-Kafka RTM benchmark. Reads from an input Kafka topic, applies a + * stateless transformation, and writes results to an output Kafka topic using + * [[RealTimeTrigger]]. After the run it reports e2e latency percentiles. + * + * The benchmark spins up a real local-cluster Spark context and a live embedded Kafka + * broker, so a single run takes several minutes. + * + * Unlike most Spark benchmarks, this one does not use `Benchmark.run()` / `addCase`: the + * metric of interest is end-to-end latency percentiles across a streaming pipeline, which + * does not fit the Best/Avg/Stdev table format. The JVM/OS/processor header that + * `Benchmark.run()` would normally emit is therefore written manually in + * `printLatenciesTable` for consistency with other benchmark result files. + * + * To run this benchmark: + * {{{ + * 1. without sbt: + * bin/spark-submit --class <this class> + * --jars <spark core test jar>,<spark sql test jar> <spark sql kafka 0-10 test jar> + * 2. build/sbt "sql-kafka-0-10/Test/runMain <this class>" + * 3. generate result: + * SPARK_GENERATE_BENCHMARK_FILES=1 build/sbt "sql-kafka-0-10/Test/runMain <this class>" + * Results will be written to: + * "connector/kafka-0-10-sql/benchmarks/RTMKafkaKafkaBenchmark-results.txt". + * }}} + * + * See `benchmarks/RTMKafkaKafkaBenchmark-results.txt` for a recorded run. + */ +object RTMKafkaKafkaBenchmark extends BenchmarkBase with Logging { + + // ----- Benchmark dimensions ----- + + // Checkpoint interval for the streaming query. 5-minute is recommended. + // Lowering it may cause more frequent checkpointing but can increase latency. + private val checkpointInterval = 5.minutes Review Comment: I think I see what you mean: in RTM, batches run for a fixed amount of time, and the batch boundary is also where progress/checkpoint-like work is committed, so this duration effectively influences the checkpoint/progress cadence. That said, since this value is passed directly to RealTimeTrigger(...), whose API names it batchDurationMs, I think calling the benchmark variable checkpointInterval is still a bit misleading. A reader may assume it is a checkpoint-specific configuration rather than the RTM batch duration. Could we name it realTimeBatchDuration or batchDuration, and keep your point in the comment? For example: ``` // Duration of each RTM batch. Since RTM commits progress/checkpoints at batch // boundaries, this also controls the effective checkpoint/progress cadence. private val realTimeBatchDuration = 5.minutes ``` This keeps the variable aligned with the public API while still explaining the RTM checkpoint/progress behavior. -- 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]
