What is the recommended approach for bootstrapping the initial load of a large dataset published to the service?

• A. Use a scratchpad dataset
• B. Use an incremental load only
• C. Use a full initial load and then incremental
• D. Use no partitioning

Answer: (C)

Establishing a complete baseline before continuing with updates is essential. Do a full initial load to populate every record in the dataset and set up the necessary structures, then switch to incremental loads to apply subsequent changes (inserts, updates, deletes) as they come. This ensures a consistent starting point, enables validation against the full snapshot, and minimizes the window where the published data could be incomplete. After the baseline is in place, ongoing changes can be published efficiently through incremental loads. Relying only on incremental updates leaves gaps if the initial state isn’t fully captured, and a scratchpad is just staging, not the published baseline.