What is a "checkpoint" in a pipeline execution and why is it useful?

• A. It is a logging mechanism.
• B. It stores a recovery point that allows resuming after a failure.
• C. It validates inputs.
• D. It allocates compute resources.

Answer: (B)

A checkpoint is a saved recovery point in a pipeline’s execution. It captures progress at a known state so that if a failure occurs later, the pipeline can resume from that point instead of starting over from the beginning. This is especially valuable for long-running data workflows, where reprocessing everything after a crash would waste time and resources. By recording which steps completed and possibly storing intermediate outputs, checkpoints enable fault-tolerant, faster recovery and allow you to re-execute only the remaining parts after addressing the issue. This contrasts with logging, which merely records events; input validation, which ensures data quality before processing; and resource allocation, which is about provisioning compute rather than preserving progress.