A company runs an asynchronous API on Amazon API Gateway with an AWS Lambda function that stores user requests in Amazon DynamoDB before dispatching them to microservices. The company provisioned as much DynamoDB throughput as budget allows but still experiences availability issues and lost requests. What should a solutions architect do to address this without impacting existing users?

The correct solution is to use Amazon SQS and Lambda to buffer writes to DynamoDB. This decouples the API Gateway from DynamoDB, allowing requests to be queued in SQS during peak loads. Lambda can then process these messages from SQS and write them to DynamoDB at a rate the database can handle, preventing lost requests and improving availability without impacting existing users. Adding throttling on API Gateway with server-side throttling limits would reject requests, impacting users. DynamoDB Accelerator (DAX) is a caching service primarily for read-heavy workloads, not for buffering writes to address throughput limitations. Creating a secondary index in DynamoDB would improve query performance for specific access patterns but would not resolve issues related to insufficient write throughput.