What is a replica set, and what do write concern and read preference control?

Short answer

A replica set is a primary plus secondaries that replicate the oplog — that is high availability and failover. Write concern is how many members must acknowledge a write. Read preference is which members are allowed to serve reads.

Why?

w: 1 acknowledges on the primary and can be rolled back if that primary dies before replicating. w: 'majority' survives failover and is the safe default for important writes. Reading from secondaries adds capacity but can return stale data because replication is asynchronous — fine for analytics, wrong for read-after-write. Sharding is a different axis: it distributes data for horizontal scale, not the same as replication.

Interview tip

Connect majority write concern to failover, not to a vague 'safer'. Mention that sharding is not how you get high availability.

Common mistake

Mixing up replica sets (copies of the same data) with sharding (slices of the data).