Which statement best describes causal consistency?

• A. All reads reflect the latest writes on every replica.
• B. Reads respect cause-effect ordering; not necessarily reflecting the most recent write.
• C. Updates propagate instantly across all replicas.
• D. The system guarantees global ordering of events but only for specific transactions.

Answer: (B)

Causal consistency ensures that reads honor the cause-and-effect relationships between operations across replicas. If one write causally precedes another, any read that observes the later write must also reflect the earlier one in the same order. But if two writes are independent (concurrent) with no causal link, a read may show either one or the other, and different replicas might reflect different orders without violating causality. This means you don’t have to see the most recent write on every replica, only the writes that causally influence what you read.

The other descriptions describe stronger guarantees: requiring the latest write on every replica implies a stronger form than causal consistency; insisting updates propagate instantly ignores network delays; insisting on a global ordering of all events corresponds to a serializability or linearizability level that goes beyond causal consistency.