A company owns an asynchronous API that is used to ingest user requests and, based on the request type, dispatch requests to the appropriate microservice for processing. The company is using Amazon API Gateway to deploy the API front end, and an AWS Lambda function that invokes Amazon DynamoDB to store user requests before dispatching them to the processing microservices.
The company provisioned as much DynamoDB throughput as its budget allows, but the company is still experiencing availability issues and is losing user requests.
What should a solutions architect do to address this issue without impacting existing users?
A. Add throttling on the API Gateway with server-side throttling limits.
B. Use DynamoDB Accelerator (DAX) and Lambda to buffer writes to DynamoDB.
C. Create a secondary index in DynamoDB for the table with the user requests.
D. Use the Amazon Simple Queue Service (Amazon SQS) queue and Lambda to buffer writes to DynamoDB.
The best solution to address this issue without impacting existing users would be Option D: Use the Amazon Simple Queue Service (Amazon SQS) queue and Lambda to buffer writes to DynamoDB.
Here's why:
- Option A: Add throttling on the API Gateway with server-side throttling limits. This option might prevent the system from being overwhelmed by too many requests at once, but it doesn't solve the problem of DynamoDB's throughput limit being reached. It might also result in user requests being denied if the throttling limit is too low.
- Option B: Use DynamoDB Accelerator (DAX) and Lambda to buffer writes to DynamoDB. DAX is a fully managed, highly available, in-memory cache for DynamoDB that can accelerate reads from DynamoDB tables by up to 10 times, even at millions of requests per second. However, DAX does not help with write throughput.
- Option C: Create a secondary index in DynamoDB for the table with the user requests. Secondary indexes can improve read performance, but they do not increase write capacity.
- Option D: Use the Amazon Simple Queue Service (Amazon SQS) queue and Lambda to buffer writes to DynamoDB. This option decouples the data ingestion process from the write process into DynamoDB. The SQS service can handle any level of throughput without losing messages or requiring other services to throttle themselves. This is likely the best solution because it addresses the issue of lost user requests due to exceeding the provisioned DynamoDB throughput.
Remember, it's always important to consider your specific use case and requirements when choosing a solution! 😊