The correct answer is: Use Amazon Simple Queue Service (Amazon SQS) with a Lambda function to buffer writes to DynamoDB.

Amazon SQS provides a highly durable and scalable message queue that decouples the incoming request flow from downstream systems. By writing user requests to SQS first, the system can absorb request spikes even if DynamoDB is throttled. An AWS Lambda function can process the SQS messages and write them to DynamoDB at a steady rate, preventing dropped requests and smoothing out throughput consumption. This allows the system to scale without affecting real-time user interactions.

Why the other options are incorrect:

Adding throttling to API Gateway would reduce the request rate and protect DynamoDB, but it directly impacts users by limiting the number of allowed requests, which results in visible API failures or rejections.

Using DynamoDB Accelerator (DAX) improves read performance by caching, but it does not address write throughput or buffering. In this use case, the issue is related to writing to DynamoDB, not reading from it.

Creating a secondary index only provides an alternative access pattern and does not improve write throughput or reduce the rate of request loss. In fact, adding indexes can slightly increase write costs and capacity consumption.

Why this question matters: This question tests your knowledge of designing resilient, decoupled serverless architectures that can handle variable workloads. Understanding how to use SQS as a buffer for write-intensive operations is critical in distributed systems design on AWS.

Tip: When you're hitting throughput limits or experiencing throttling with DynamoDB, consider SQS or Kinesis to decouple producers from consumers. This provides elasticity and protects your data from being dropped during peak traffic periods.

Full AWS Practitioner Certification Question