A podcast platform detects low engagement by analyzing a 10-minute sliding window of user events. You must ingest events and transform the most recent 10 minutes of data before running inference with minimal operational overhead. Which design is best?

The correct option leverages Kinesis Data Streams for real-time ingestion and Kinesis Data Analytics for continuous, window-based processing. Kinesis Data Analytics is specifically designed for SQL-based or Apache Flink-based analysis on streaming data, making it ideal for defining a 10-minute sliding window and transforming events in real time with minimal operational overhead. Incorrect options: Using Amazon MSK (Managed Streaming for Apache Kafka) for ingestion is viable, but Kinesis Data Analytics is a more streamlined solution for windowed stream processing compared to integrating MSK with a separate processing engine for this specific task. Delivering to S3 with Kinesis Data Firehose introduces latency and storage overhead, making it unsuitable for real-time, windowed analysis. AWS Lambda is good for event-driven processing but less efficient for continuous, stateful windowing operations compared to Kinesis Data Analytics. Using AWS Lambda with MSK for windowed processing would require more custom development and state management, increasing operational overhead compared to Kinesis Data Analytics' built-in capabilities.