Enhance your career, get your certificate as a Data Streaming Engineer | Get your Certificate
Confluent Developer Blog
Flink SQL Evolution: Handling Custom CDC with FROM_CHANGELOG and TO_CHANGELOG
How the new FROM_CHANGELOG and TO_CHANGELOG built-in functions let Flink SQL read custom CDC formats and turn updating streams back into append-only ones.
Absolutely Everything You Always Wanted to Know About Watermarks in Apache Flink - Part 2: Confluent Cloud for Apache Flink
Confluent Cloud for Apache Flink watermarks deep-dive. Differences and extensions to Apache Flink: default watermarks, idleness, alignment, late events handling, watermark observability.
Absolutely Everything You Always Wanted to Know About Watermarks in Apache Flink - Part 1: Apache Flink
Deep-dive reference on Apache Flink watermarks: generation, propagation, idle partition detection, alignment, and why lateness is non-deterministic.
A Fast Path for Fixed-Length Lists in Parquet
How Hardwood detects fixed-length lists in Parquet's encoded definition and repetition levels, bypassing Dremel reconstruction for a 1.1×–3.9× parse speed-up.
Kafka Queues for Quarkus
Learn how to use Queues for Apache Kafka (KIP-932) with the Quarkus framework, including share group configuration, explicit message acknowledgment, and cooperative consumption from a single partition.
How we cut Flink OOMKills by 91.2%: Zombie block cache, phantom CPUs (and some bonus AI Learnings)
How we traced a native-memory leak in Flink's RocksDB block cache and a jemalloc fragmentation bug, cutting OOMKilled Task Managers by 91.2% — plus what running the investigation with Claude taught us.
