Yesterday's Kinesis Data Streams are powerful but labor-intensive. You calculate shard counts, write consumer code (KCL/Lambda), manage checkpoints, and implement your own S3 loading logic. You often think, "I just wish streaming data would load automatically to S3 or Redshift." The answer is Amazon Kinesis Data Firehose (now called Amazon Data Firehose).
Firehose in one sentence is "a fully managed data delivery pipe." Producers push data in, Firehose buffers it automatically, optionally transforms, compresses, and converts formats, then delivers it to the destination. No shards, no consumer code, no scaling — it all just works.
The exam frequently contrasts these two