A company collects data from thousands of remote devices using a RESTful API running on an Amazon EC2 instance. The EC2 instance receives raw data, transforms it, and stores it in Amazon S3. The number of devices will soon grow into the millions. The company needs a highly scalable solution with minimal operational overhead. Which combination of steps should a solutions architect take? (Choose two.)

The current architecture, relying on a single EC2 instance for both API and processing, will not scale to millions of devices. Using Amazon API Gateway provides a highly scalable, managed entry point for the RESTful API, eliminating the need to manage EC2 instances for this purpose. Sending data to an Amazon Kinesis data stream allows for real-time, scalable ingestion of millions of data points. Kinesis Data Firehose can then reliably deliver this streaming data to Amazon S3 with minimal operational overhead. Once data is in S3, AWS Glue is an excellent choice for serverless, scalable data transformation and processing, further reducing operational burden compared to managing EC2 instances for this task. Adding more EC2 instances for the API or processing would increase operational overhead and not scale as efficiently as managed services. Using Amazon SQS with EC2 instances for processing would still require managing EC2 instances and might introduce latency for real-time processing needs. Route 53 is a DNS service and doesn't directly address the scalability of the API or data processing.