Data lakehouses often run multiple concurrent writers—streaming ingestion, batch ETL, maintenance jobs. The default approach, Optimistic Concurrency Control (OCC), assumes conflicts are rare and handles them through retries. That assumption breaks down in increasingly common scenarios, such as running maintenance batch jobs on tables receiving streaming writes.
Maximizing Throughput with Apache Hudi NBCC: Stop Retrying, Start Scaling
calendar_today
December 16, 2025
domain
apache-hudi