A Firehose stream writes clickstream events to S3, and consumers complain about millions of tiny objects. Which setting should be tuned so that records accumulate before each delivery?
Show answer and explanation
Answer: A
Firehose accumulates incoming records in an internal buffer and delivers them when either the buffered size or the elapsed time threshold is reached, whichever comes first. Raising the size and interval produces fewer, larger objects, which improves downstream scan efficiency and reduces per-object overhead. Compression changes how an object is encoded, the delimiter separates records inside a delivered file, and the IAM role governs permissions, so none of them controls how many objects are produced.
Why the other options are wrong
- B. Compression shrinks each delivered object but does not change how many objects the stream creates.
- C. The record delimiter marks boundaries inside a file and has no effect on delivery frequency or object count.
- D. The IAM role determines what Firehose may access; it cannot alter buffering behaviour.
Exam tip. Firehose flushes on whichever comes first: buffer size or buffer interval.