AWS Certified Solutions Architect – Associate (SAA-C02) — Question 524

A company has an application that collects data from loT sensors on automobiles. The data is streamed and stored in Amazon S3 through Amazon Kinesis Data
Firehose. The data produces trillions of S3 objects each year. Each morning, the company uses the data from the previous 30 days to retrain a suite of machine learning (ML) models.
Four times each year, the company uses the data from the previous 12 months to perform analysis and train other ML models. The data must be available with minimal delay for up to 1 year. After 1 year, the data must be retained for archival purposes.
Which storage solution meets these requirements MOST cost-effectively?

Answer options

Correct answer: D

Explanation

Because the application generates trillions of objects, using S3 Intelligent-Tiering (Options A and B) is highly cost-prohibitive due to its per-object monitoring and automation charges. Transitioning to S3 Standard-IA immediately (Option C) is also inefficient because the daily ML retraining requires frequent access to the first 30 days of data, which would trigger high retrieval fees. Option D is the most cost-effective approach because it uses S3 Standard for the active first 30 days, transitions to S3 Standard-IA for the quarterly retrievals, and archives to S3 Glacier Deep Archive after 1 year.