A company's asynchronous API, deployed via Amazon API Gateway, and a Lambda function interfacing with Amazon DynamoDB, is experiencing availability issues despite budget-constrained throughput provisioning. How can a solutions architect mitigate request loss without affecting current users?