Neu in Confluent Cloud: Daten & Pipelines für KI-fähiges Streaming zugänglich machen | Mehr erfahren
A few months ago, we announced the major release of Apache Kafka 0.9, which added several new features like Security, Kafka Connect, the new Java consumer and also critical bug fixes. Today, I am pleased to announce the availability of the first patch release of the Kafka 0.9 series: the 0.9.0.1 release of Apache Kafka and the corresponding patch release of the Confluent Platform 2.0.1. The 0.9.0.1 release fixes a total of 68 issues, out which 45 are bug fixes and 20 are general improvements.
We recommend all users upgrade to this release by bumping up the version of your Kafka dependencies to 0.9.0.1 (Confluent Platform users: 0.9.0.1-cp1) in your applications and updating the binary installations on server machines.
Here is a quick overview of the notable Kafka-related bug fixes in the release, grouped by the affected functionality. You can review the complete list of issues fixed in the release notes.
According to git shortlog, 42 people contributed to this release –
Adam Kunicki, Alex Sherwin, Ashish Singh, Ben Stopford, Binlei Xu, David Jacot, Denise Fernandez, Dmitry Stratiychuk, Dong Lin, Edward Ribeiro, Eno Thereska, Ewen Cheslack-Postava, Geoff Anderson, Grant Henke, Guozhang Wang, Gwen Shapira, Ismael Juma, Jaikiran Pai, James Cheng, Jason Gustafson, Jay Kreps,
Jesse Anderson, Jiangjie Qin, Jin Xing, Jun Rao, Kim Christensen, Kishore Senji, Luciano Afranllie, Magnus Edenhill, Maksim Logvinenko, Mayuresh Gharat,
Michael Blume, Piotr Szwed, Praveen Devarao, Rajini Sivaram, Sasaki Toru, Tao Xiao, Tom Graves, Tomasz Nurkiewicz, Vahid Hashemian, Xin Wang, Yifan Ying
The easiest way to get started with or upgrade Kafka is by downloading Confluent Platform. The 2.0.1 release of Confluent Platform is 100% open-source and includes Apache Kafka 0.9.0.1 along with tools that you need to get started with Kafka. Learn more about it by reading the details in the Confluent Platform 2.0.1 documentation or download it to give it a spin.
Confluent Platform 2.0.1 is backed by our subscription support, and we also offer expert training and technical consulting to help get your organization started.
As always, we are happy to hear your feedback. Please post your questions and suggestions to the public Confluent Platform mailing list.
Batch CDPs can't capture user intent as it forms. By the time a nightly sync runs, the moment is gone. This guide covers the streaming architecture behind real-time personalization, from sub-100ms ad bidding to cross-channel orchestration, with recommendation patterns built on Kafka and Flink.
Separate batch and streaming pipelines for ML features cause training-serving skew. DoorDash measured a 35.7% feature mismatch in their dual setup. This guide covers a unified kappa architecture using Flink to compute features once for both training and serving, plus a 2026 tooling comparison.