An application uses Amazon API Gateway and an AWS Lambda function to fetch data from Amazon DynamoDB. Users report intermittent high latency. The team's investigation shows the API's Lambda function is frequently throttled whenever other Lambda functions have higher invocation rates. The company wants the API's Lambda to be isolated from other functions in the most cost-effective way. What should they do?

Configuring reserved concurrency for the Lambda function dedicates a specific number of concurrent executions to that function, preventing other Lambda functions from consuming those resources. This isolates the API's Lambda from throttling due to high invocation rates from other functions, making it the most cost-effective solution as you only pay for the reserved concurrency when it's used. Increasing DynamoDB RCUs would address database bottlenecks, not Lambda throttling. Provisioned concurrency keeps functions warm, reducing cold start latency, but doesn't prevent throttling from other functions. Increasing timeout and memory might help function performance but won't prevent throttling if the overall account concurrency limit is reached by other functions.