An application running on Amazon EC2 calls the DynamoDB REST API. Periodically, writes fail with ProvisionedThroughputExceededException. Which changes will most cost-effectively reduce these errors? (Choose two.)

Exponential backoff is a standard retry mechanism where an application waits for progressively longer periods between retries of failed requests. This prevents overwhelming the service and allows temporary capacity issues to resolve, making it a cost-effective solution for ProvisionedThroughputExceededException errors without incurring additional AWS costs. AWS SDKs automatically incorporate best practices, including exponential backoff and intelligent retry logic, which significantly improves resilience and reduces the likelihood of these errors. Using the SDKs is a fundamental best practice for interacting with AWS services. Increasing provisioned capacity would resolve the errors but is not the most cost-effective as it directly increases billing. DAX is for read-heavy workloads and won't address write capacity issues. Splitting tables adds complexity and management overhead, and doesn't inherently solve the underlying throughput issue without also increasing capacity on the new tables.