Auto Loader discovers new files and tracks progress. Its reader uses format('cloudFiles') with cloudFiles.format naming the actual input format. A schema location stores schema information; a streaming checkpoint records progress and state. Those locations have different jobs and should remain stable for the same logical pipeline.
The availableNow trigger processes currently available data and stops. That makes it useful for scheduled incremental loads. The simulation provides a finite landing directory and remembers processed input within a checkpoint. It does not emulate cloud notifications or continuously running executors. Do not reuse one checkpoint for unrelated sources. Deleting it can cause replay, and a downstream append may then duplicate records. Schema evolution also needs a policy: new columns can be accepted deliberately, while incompatible values still need validation and a recovery plan.