Sharding
Problem
Your database contains so much data or traffic that one machine can no longer handle it.
Simply making another full copy doesn't divide the workload.
Solution — Sharding
Partition data across machines.
Customers A–F → Shard 1 Customers G–M → Shard 2 Customers N–Z → Shard 3
Each node owns part of the dataset.
Core idea: sharding scales by dividing data across independent storage nodes.