AWS Certified Developer Associate DVA-C02 Practice Question

Implementing retries with exponential backoff is the correct strategy for dealing with exceeded throughput limits in a database like DynamoDB. When a ProvisionedThroughputExceededException occurs, it means the allowed capacity has been temporarily exceeded. To handle this, the application should implement a delay before trying the same request again. The delay should gradually increase with each attempt (exponential backoff), and ideally, there should be a variation in the retry intervals (jitter) to prevent synchronized retries from different instances, which could continue to overwhelm the database. This pattern helps to distribute the load more evenly and increases the likelihood of successful requests without continuously hitting the throughput limit. Directly retrying without any delay or backoff will likely result in repeated exceptions because the database throughput may not have had the chance to recover. Increasing the provisioned throughput immediately is a reactive measure that does not solve the immediate problem and may lead to higher costs, while the application cannot automatically switch a table to on-demand capacity in response to exceptions.