On 16 September 2026, Salesforce went down across six countries for about ten and a half hours. The public explanation is one sentence long, and it describes the most common cascading failure in backend systems: requests piling up behind one slow internal service until no workers were left. Here is why timeouts alone do not save you, what bulkheads cost, and why recovery took far longer than the failure.
A five-step order flow dies after step three: the card is charged, the stock is held, and the process that knew what came next is gone. Sagas, outboxes, and idempotency keys are how we've survived that crash for a decade — by hand-building a state machine every single time. Durable execution flips the model: write the flow as straight-line code, and let an engine replay it back to the exact line where it died.