A marketing team must score 40 million customer rows once a week and write the scores back to storage. No one needs answers in real time. Which option is the most economical?
Show answer and explanation
Answer: B
Batch prediction processes large volumes of records as a managed job that reads from and writes to storage, and it bills for the job's compute rather than an always-on endpoint. That suits a weekly scoring run with no latency requirement. Always-on endpoints waste capacity between runs, streaming adds complexity for no benefit, and manual entry cannot scale.
Why the other options are wrong
- A. An endpoint kept running all week pays for idle capacity, and sending 40 million single requests is slower than a bulk job.
- C. Per-row streaming serves a need that does not exist here, and it adds infrastructure cost and complexity for no latency gain.
- D. Manual entry cannot process forty million rows in any practical timeframe, and it introduces errors and leaves no audit trail.