A company operates an IoT solution with millions of home sensors in the United States. Sensors use MQTT to send data to a custom MQTT broker on a single EC2 instance. Sensors connect to the broker at iot.example.com (Route 53). The broker writes data to DynamoDB. The broker has been overloaded multiple times, causing data loss. Which solution will improve reliability and meet the requirements?

The correct solution leverages AWS IoT Core, a fully managed service designed for IoT devices, to handle millions of connections and messages reliably. By configuring a custom domain and updating Route 53 to point to the AWS IoT Core Data-ATS endpoint, the existing sensor configuration (connecting to iot.example.com) can be maintained with minimal changes. An AWS IoT rule then efficiently routes the ingested data to DynamoDB, replacing the custom broker's data storage function. This eliminates the single point of failure and scalability issues of the custom EC2-based MQTT broker. The other options are less suitable: An ALB does not support MQTT directly, and while an NLB could be used, it still relies on managing custom MQTT brokers, which introduces operational overhead and doesn't fully address the scalability and reliability needs as effectively as IoT Core. AWS IoT Greengrass is for edge device processing, not a centralized ingestion point for millions of remote sensors.