Investigating retry patterns in CloudWatch

Your team at ReliableApp Co has a Lambda function called payment-processor that handles payment notifications asynchronously. The function occasionally fails, and your operations team has configured a dead-letter queue (DLQ) called payment-processor-dlq to capture events that cannot be processed. A colleague asks you how Lambda's retry behavior works with the DLQ.

By default, how many times does Lambda retry a failed asynchronous invocation before sending the event to the DLQ?