Incremental Loading
Processing only the data that has changed since the last run, rather than reprocessing everything.
Mechanism
- Watermarking: We store a
last_run.jsoncontaining the timestamp of the last successful extraction. - Filtering: The extraction phase only requests data
updated_at > last_run_timestamp. - Upserting: During the load phase, we handle duplicates using primary keys to ensure idempotency.
This ensures the pipeline is fast and cost-effective.