Compare

Side-by-side on the decisions that recur: index vs scan, normalize vs denormalize, optimistic vs pessimistic, partition vs shard, and more — with when to choose each.

Read replicaShard
ScalesReadsWrites (and storage)
DataFull copy of everythingA slice of the keyspace
ConsistencyReplication lag on readsCross-shard operations lose atomicity
FailoverPromote a replicaPer-shard concerns
ComplexityLow — add and route readsHigh — routing, rebalancing, cross-shard logic
Choose this whenA read-heavy workload (most are) — near-linear read capacity and a failover target.A write-bound workload that has outgrown one machine — replicas cannot help writes.