blob: 17b0fece876056c7c1ca20518e3e3d209111a1b5 [file] [log] [blame]
/*
* 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.loadtests;
import static java.lang.String.format;
import java.io.IOException;
import java.math.BigInteger;
import java.util.Optional;
import org.apache.beam.sdk.io.synthetic.SyntheticStep;
import org.apache.beam.sdk.options.Default;
import org.apache.beam.sdk.options.Description;
import org.apache.beam.sdk.testutils.metrics.ByteMonitor;
import org.apache.beam.sdk.testutils.metrics.TimeMonitor;
import org.apache.beam.sdk.transforms.Combine;
import org.apache.beam.sdk.transforms.Count;
import org.apache.beam.sdk.transforms.MapElements;
import org.apache.beam.sdk.transforms.Mean;
import org.apache.beam.sdk.transforms.PTransform;
import org.apache.beam.sdk.transforms.ParDo;
import org.apache.beam.sdk.transforms.SimpleFunction;
import org.apache.beam.sdk.transforms.Sum;
import org.apache.beam.sdk.transforms.Top;
import org.apache.beam.sdk.values.KV;
import org.apache.beam.sdk.values.PCollection;
import org.apache.beam.vendor.guava.v26_0_jre.com.google.common.base.Preconditions;
/**
* Load test for {@link ParDo} operation.
*
* <p>The purpose of this test is to measure {@link Combine}'s behaviour in stressful conditions. It
* uses synthetic sources and {@link SyntheticStep} which both can be parametrized to generate keys
* and values of various size, impose delay (sleep or cpu burnout) in various moments during the
* pipeline execution and provide some other performance challenges.
*
* <p>You can choose between multiple combine modes to test per key combine operations ({@link
* CombinerType}).
*
* <p>To run it manually, use the following command:
*
* <pre>
* ./gradlew :sdks:java:testing:load-tests:run -PloadTest.args='
* --fanout=1
* --perKeyCombinerType=TOP_LARGEST
* --topCount=10
* --sourceOptions={"numRecords":1000,...}
* --stepOptions={"outputRecordsPerInputRecord":2...}'
* -PloadTest.mainClass="org.apache.beam.sdk.loadtests.CombineLoadTest"
* </pre>
*/
public class CombineLoadTest extends LoadTest<CombineLoadTest.Options> {
private static final String METRICS_NAMESPACE = "combine";
/** Enumerates per-key combiners available in the test. */
public enum CombinerType {
TOP_LARGEST,
MEAN,
SUM,
COUNT
}
/** Pipeline options specific for this test. */
public interface Options extends LoadTestOptions {
@Description("Number consequent of ParDo operations (SyntheticSteps) to be performed.")
@Default.Integer(1)
Integer getNumberOfCounterOperations();
void setNumberOfCounterOperations(Integer count);
@Description("The number of Combine operations to perform in parallel.")
@Default.Integer(1)
Integer getFanout();
void setFanout(Integer fanout);
@Description("Per key combiner type.")
@Default.Enum("MEAN")
CombinerType getPerKeyCombiner();
void setPerKeyCombiner(CombinerType combinerType);
@Description("Number of top results to combine (if applicable).")
Integer getTopCount();
void setTopCount(Integer topCount);
@Description("Number of reiterations over the values to perform.")
@Default.Integer(1)
Integer getIterations();
void setIterations(Integer iterations);
}
private CombineLoadTest(String[] args) throws IOException {
super(args, Options.class, METRICS_NAMESPACE);
}
@Override
protected void loadTest() throws IOException {
Optional<SyntheticStep> syntheticStep = createStep(options.getStepOptions());
PCollection<KV<byte[], byte[]>> input =
pipeline
.apply("Read input", readFromSource(sourceOptions))
.apply(
"Collect start time metric",
ParDo.of(new TimeMonitor<>(METRICS_NAMESPACE, "runtime")))
.apply(
"Collect metrics",
ParDo.of(new ByteMonitor(METRICS_NAMESPACE, "totalBytes.count")));
input = applyWindowing(input);
for (int i = 0; i < options.getFanout(); i++) {
applyStepIfPresent(input, format("Step: %d", i), syntheticStep)
.apply(format("Convert to Long: %d", i), MapElements.via(new ByteValueToLong()))
.apply(format("Combine: %d", i), getPerKeyCombiner(options.getPerKeyCombiner()))
.apply(
"Collect end time metric", ParDo.of(new TimeMonitor<>(METRICS_NAMESPACE, "runtime")));
}
}
public PTransform<PCollection<KV<byte[], Long>>, ? extends PCollection> getPerKeyCombiner(
CombinerType combinerType) {
switch (combinerType) {
case MEAN:
return Mean.perKey();
case TOP_LARGEST:
Preconditions.checkArgument(
options.getTopCount() != null,
"You should set \"--topCount\" option to use TOP combiners.");
return Top.largestPerKey(options.getTopCount());
case SUM:
return Sum.longsPerKey();
case COUNT:
return Count.perKey();
default:
throw new IllegalArgumentException("No such combiner!");
}
}
private static class ByteValueToLong
extends SimpleFunction<KV<byte[], byte[]>, KV<byte[], Long>> {
@Override
public KV<byte[], Long> apply(KV<byte[], byte[]> input) {
return KV.of(input.getKey(), new BigInteger(input.getValue()).longValue());
}
}
public static void main(String[] args) throws IOException {
new CombineLoadTest(args).run();
}
}