An application server crashes, but a new server instance can resume processing using data from...

Stateless design means that individual application server instances do not retain session-specific or transactional state locally; instead, all state required to continue processing is kept in a centralized, shared repository (such as a database or distributed cache) accessible to any server instance. This is precisely what allows a brand-new server instance to seamlessly resume processing after the original instance crashes, since the new instance simply reads the necessary state from the shared repository rather than needing to recover state that was trapped on the failed server, which is the defining characteristic and benefit of stateless architecture for resiliency and failover. Graceful degradation (A) describes continuing to operate with reduced functionality when a component fails, a different concept from seamlessly resuming full processing on a replacement instance using externally stored state. Database rollback (B) refers to reverting a transaction or dataset to a previous consistent state, typically in response to an error or failed transaction, which is unrelated to how a new server instance picks up processing after a crash. Circuit breakers (D) are a pattern for detecting and short-circuiting calls to a failing downstream dependency to prevent cascading failure, a different resiliency mechanism entirely from state externalization. Stateless design correctly explains this scenario.

Reference topic: Fault Tolerance Techniques - Stateless Application Design.