Retries

A retry is trying a failed request again, on the theory that many failures are brief, like a dropped packet or a server restarting, and the second attempt will work. Done naively, retries make outages worse: if a service is struggling and every client immediately retries several times, load multiplies exactly when capacity is lowest, a retry storm. So good retries use exponential backoff, waiting longer after each failure, plus jitter, a random spread so clients that failed together don’t all come back at the same instant. You also cap the number of attempts, only retry errors that are likely temporary, and make sure the operation is idempotent so repeating it can’t, say, charge a card twice.