What should the solutions architect recommend to improve the customer experience?
Implement retry logic with exponential backoff and irregular variation in the client application. Ensure that the errors are caught and handled with descriptive error messages.
Implement API throttling through a usage plan at the API Gateway level. Ensure that the client application handles code 429 replies without error.
Turn on API caching to enhance responsiveness for the production stage. Run 10-minute load tests. Verify that the cache capacity is appropriate for the workload.
Implement reserved concurrency at the Lambda function level to provide the resources that are needed during sudden increases in traffic.
Explanations:
While exponential backoff helps with retries in the client, it does not directly address rate limiting or prevent error accumulation on the API Gateway. The solution should control traffic at the API level instead.
API throttling with usage plans at the API Gateway level can effectively limit the requests from a high-traffic client, reducing errors due to request overload. It also ensures clients handle HTTP 429 responses gracefully.
API caching improves read performance but does not help with PUT requests, as caching is generally used for GET requests. PUT requests still directly access the backend resources.
Reserved concurrency limits the Lambda function’s total concurrent executions but does not manage client-side traffic or address issues stemming from a high volume of requests from specific clients.
Based on what I know, the answer is:
Implement API throttling through a usage plan at the API Gateway level. Ensure that the client application handles code 429 replies without error.