A vehicle manufacturer collects telemetry via MQTT from many vehicles. Data is processed every 5 minutes and copied to on-premises storage for anomaly detection by custom apps. Because on-premises storage can't scale to peaks, data is lost. Which AWS-based solution with the least operational overhead solves the scaling and analytics needs?

The correct solution leverages AWS IoT Core for scalable ingestion of MQTT data, directly addressing the initial scaling problem. Kinesis Data Firehose then reliably delivers this data to Amazon S3, providing a cost-effective and highly scalable storage solution for the 5-minute processing window. Kinesis Data Analytics offers a serverless platform to run real-time anomaly detection on the streaming data with minimal operational overhead, fulfilling the analytics requirement. Incorrect options: AWS IoT Greengrass is for edge processing, not primary cloud ingestion, and Amazon MSK adds operational complexity compared to serverless options. SageMaker requires more setup for a pretrained model than Kinesis Data Analytics for streaming anomaly detection. AWS IoT FleetWise is for collecting, standardizing, and transforming vehicle data, not just ingestion, and Kinesis Data Streams would require more management than Firehose for direct S3 delivery. AWS Glue machine learning transforms are typically for batch processing, not real-time anomaly detection on streaming data. Amazon MQ for RabbitMQ is a managed message broker, but IoT Core is specifically designed for device connectivity and MQTT at scale. Amazon Lookout for Metrics is a good anomaly detection service, but the overall architecture with MQ is less optimized for device telemetry compared to IoT Core and Kinesis.