Is the process manual? I want to embed the approach in an automation workflow. For time travel, we can get it. But I don't see it as viable to handle, say, 50–60 GB of data.

Automating a manual data pipeline involves identifying repeatable steps, scripting them, and orchestrating with a workflow engine like Airflow or Prefect. For large volumes (50–60 GB), you should use incremental or partitioned loads rather than full refreshes. Time‑travel (data versioning) can be achieved with a CDC source or a time‑stamped column and a versioned table in a data lake. To keep the pipeline viable, use parallelism, chunking, and efficient storage formats (Parquet/ORC). Add idempotency checks and retries to handle failures. Finally, monitor resource usage and set alerts so the automation remains sustainable.