Gunnar Morling, Java Champion and technologist at Confluent, discusses high-performance Java development across several topics. He reflects on the One Billion Row Challenge and how modern JDK versions (17+) provide out-of-the-box performance gains via compact object headers and improved ZGC. He explains durable execution engines and his prototype Persistasaurus, which uses SQLite to make long-running Java workflows resumable and recoverable. The main focus is Hardwood, a new zero-dependency Apache Parquet parser for Java that uses page-level parallelism with virtual threads and adaptive thread balancing to maximize CPU utilization. Hardwood was built AI-natively using Claude Code, with Gunnar emphasizing the importance of human oversight, design documents, and CLAUDE.md guidelines to maintain code quality. The conversation also touches on AI's bimodal impact on developers — a productivity booster for experienced engineers but potentially disruptive for newcomers.
Table of contents
TranscriptReflecting on the One Billion Row Challenge's Unexpected Impact [ 02:32 ]JDK 17 as the New Baseline Provides More Performance Out of the Box [ 04:50 ]Durable Execution in Java: Writing Resumable, Recoverable Workflows in Plain Code [ 08:59 ]Hardwood: Building a Fast, Zero-Dependencies Parquet Parser in Java [ 14:52 ]Hardwood's Architecture: Parallelisation and Performance Optimisation [ 20:25 ]AI-Native Development: The Massive Productivity Booster that Needs Human Oversight [ 27:33 ]The Future of Programming: AI, Cost, and Maintaining Developer Skills [ 36:52 ]Contributing to Hardwood Community [ 40:05 ]About the AuthorSort: