HOW JD.COM BUILT A LAKESTREAM WITH APACHE FLUSS AND APACHE HUDI

Summary from the TLDR Data newsletter

JD replaced parallel streaming and batch training-data pipelines with a staged migration: write once to Fluss, tier to Hudi, and union fresh plus historical reads. The difficult part is consistency: workers coordinate a single Hudi commit per tiering round and report matched stream offsets only after it completes. That pattern enables rollback-safe catch-up, but leaves schema evolution and dynamic bucketing unresolved.