anjy7 commented on code in PR #22669: URL: https://github.com/apache/kafka/pull/22669#discussion_r3572139593
########## jmh-benchmarks/src/main/java/org/apache/kafka/jmh/raft/ElectionBenchmarks.java: ########## @@ -0,0 +1,78 @@ +/* + * 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.kafka.jmh.raft; + +import org.apache.kafka.raft.RaftClientBenchmarkContext; +import org.apache.kafka.raft.RaftClientTestContext; + +import org.openjdk.jmh.annotations.Benchmark; +import org.openjdk.jmh.annotations.BenchmarkMode; +import org.openjdk.jmh.annotations.Fork; +import org.openjdk.jmh.annotations.Level; +import org.openjdk.jmh.annotations.Measurement; +import org.openjdk.jmh.annotations.Mode; +import org.openjdk.jmh.annotations.OutputTimeUnit; +import org.openjdk.jmh.annotations.Param; +import org.openjdk.jmh.annotations.Scope; +import org.openjdk.jmh.annotations.Setup; +import org.openjdk.jmh.annotations.State; +import org.openjdk.jmh.annotations.Warmup; + +import java.util.Optional; +import java.util.concurrent.TimeUnit; + +/** + * Benchmarks for the leader-election path. The outer class is intentionally not a JMH {@code @State}: + * each benchmark declares the starting state it needs as a nested {@code @State} parameter, so + * different election scenarios (e.g. a future Prospective or Candidate start) can have their own + * setup without forcing a single shared {@code @Setup} on the whole class. + */ +@BenchmarkMode(Mode.SingleShotTime) Review Comment: Good questions! In this particular benchmark, the `setup cost= construct and initialize a fresh KafkaRaftClient + its mocks + write the initial quorum-state file`. Leader election is a state consuming transition i.e once we elect a leader in that particular benchmark invocation, we need to build a fresh context for the next invocation. Using `Singleshot` we can exclude this setup cost ( done before every invocation ) from the measurement, as it will do the setup outside the timed region. We could think of this as a throughput measurement too. Right now, we get the`time per election` ( without the setup noise ). If after the refactor, the election gets 2x more expensive, it would essentially mean the throughput is halved. I don't think a separate `Mode.Throughput` would be a good idea for this scenario. `Mode.Throughput` (ops/second) and `Mode.AverageTime` (nanosecond/op) are pretty much the same. They're reciprocals. These are good for "steady" state scenarios where we don't need to reset the state. For eg, a leader handling FETCH from a caught up follower. We can run this case millions of times over a time interval without the need to reset the context. -- 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]
