How do you migrate an existing Parquet-based data lake to Apache Hudi?

Hudi's bootstrap feature is purpose-built for this migration, letting you bring existing Parquet data under Hudi's management without necessarily rewriting every byte immediately.

1. Choose a bootstrap mode — metadata-only bootstrap keeps the original Parquet files as-is and just generates Hudi's metadata/index pointing at them (fast, minimal storage cost); full-record bootstrap actually rewrites the data into Hudi's file layout upfront (slower, but avoids any dependency on the original files afterward).
2. Define the HoodieKey — choose record key and partition path fields that match how the existing data is already organized, so the migration doesn't require reshuffling files across partitions.
3. Run the bootstrap job, which registers the existing files as the table's initial file slices and writes the first commit instant to the timeline.
4. Point downstream engines at the new Hudi table and validate row counts and sample queries match the original dataset.
5. Switch ingestion to Hudi's upsert/insert APIs going forward, so all new changes are properly tracked on the timeline from that point on.