Study Databricks DA-ASSOC Uploads and Auto Loader: key concepts, common traps, and exam decision cues.
The exam does not expect deep ingestion engineering, but it does expect you to pick a reasonable data-access path. The best answer usually depends on whether the need is small and manual, ongoing and repeatable, externally shared, or discovered through a Databricks marketplace surface.
| Need | Better choice |
|---|---|
| quick manual file brought in through the UI | workspace upload |
| ongoing arrival of files in cloud storage | Auto Loader |
| governed sharing with external systems without raw-copy handoff | Delta Sharing |
| discover and use external data products through Databricks | Marketplace |
| API-driven intake | platform-driven ingestion path controlled through APIs |
| If the stem says… | Strong reading |
|---|---|
| “analyst uploads a file through the UI” | workspace upload |
| “new files arrive continuously” | Auto Loader |
| “share data externally in a governed way” | Delta Sharing |
| “find and access external datasets through Databricks” | Marketplace |
| Trap | Better rule |
|---|---|
| choosing workspace upload for a repeated pipeline | manual upload fits small ad hoc work, not durable ingestion |
| treating Delta Sharing like file upload | Delta Sharing is governed sharing, not manual copy |
| mixing Marketplace with Auto Loader | Marketplace is for finding assets; Auto Loader is for ingesting arriving files |
Import questions usually reward matching the ingest method to the source pattern. If files arrive continuously in cloud storage, think Auto Loader. If the requirement is governed external data distribution, think Delta Sharing. The exam often tests whether you confuse ingestion into Databricks with sharing data out of Databricks.