https://mariano-gonzalez.com/tags/benchmarks/ Recent content in Benchmarks on Mariano Gonzalez Hugo en-us <p>Copyright © 2026 Mariano Gonzalez</p><p xmlns:dct='http://purl.org/dc/terms/' xmlns:cc='http://creativecommons.org/ns#' class='license-text'><a rel='cc:attributionURL' property='dct:title' href='https://github.com/eschizoid/eschizoid.github.io/tree/main/content'>Content</a> licensed under <a rel='license' href='https://creativecommons.org/licenses/by/4.0'>CC BY 4.0</a></p> Sat, 16 May 2026 00:00:00 +0000 https://mariano-gonzalez.com/posts/post-5/ Sat, 16 May 2026 00:00:00 +0000 https://mariano-gonzalez.com/posts/post-5/ <p>If you write a library that &ldquo;consumes Kafka in parallel,&rdquo; eventually someone asks &ldquo;compared to what?&rdquo;</p> <p>For KPipe that question had a partial answer. Up to 1.12 the benchmark compared KPipe to the Confluent Parallel Consumer only, at 10,000 records per invocation, with no per-record workload. One competitor, one workload, one number.</p> <p>This post is about the upgrade and the first numbers from it. KPipe now has a four-runtime competitive bench: <strong>KPipe</strong>, <strong>Confluent Parallel Consumer</strong>, <strong>Reactor Kafka</strong>, and a hand-rolled <strong>raw <code>KafkaConsumer</code> + virtual-thread executor</strong> baseline. Three workload regimes. Real Kafka broker (Testcontainers, not in-process). JMH-published scores, GC profiler, raw JSON committed alongside this post in <a href="https://github.com/eschizoid/kpipe/tree/main/benchmarks/results"><code>benchmarks/results/</code></a>.</p>