A company hosts an application that uploads files to an Amazon S3 bucket. After upload, files are processed to extract metadata, which takes less than 5 seconds. Upload volume varies from a few files per hour to hundreds of concurrent uploads. Design a cost-effective architecture to meet these requirements. What should the solutions architect recommend?

The correct solution is to configure an object-created event notification within the S3 bucket to invoke an AWS Lambda function. This is a cost-effective, serverless approach that scales automatically with varying upload volumes. S3 event notifications directly trigger the Lambda function when a new file is uploaded, eliminating the need for polling or managing servers. Lambda is ideal for short-duration tasks like metadata extraction (under 5 seconds) and is billed per invocation and duration, making it cost-efficient for intermittent workloads. Incorrect options: CloudTrail logs API calls but doesn't process files. AppSync is for building GraphQL APIs, not for S3 file processing. Kinesis Data Streams is for real-time streaming data, which is overkill and more expensive for simple S3 object creation events. SNS is a messaging service. While it can trigger Lambda, S3's direct event notification to Lambda is more direct and efficient for this specific use case.