Deterministic database
A database where the same set of transactions, run in the same order, is guaranteed to give exactly the same result on every replica — no random or time-dependent deviations. The property that lets replicas stay in sync without coordinating the outcome.
Idea: if the execution is fully deterministic, it's enough for all replicas to agree on the order of the input (the transactions); the result follows automatically and identically everywhere. This dramatically simplifies replication (send just the transaction stream, not the results), eliminates the need for distributed commit coordination, and makes recovery simple (replay the stream). Requirement: remove non-determinism — NOW(), RANDOM(), unordered concurrency must be handled/forbidden or made deterministic. Calvin is the well-known example; the same principle drives event sourcing, state machine replication (Raft applies a deterministic log), and blockchains' consensus. A powerful way to trade coordination for a predetermined order. Related to Calvin and event sourcing.