Range partitioning: Keys A-M on node , N-Z on node . Good for range queries. Risk of hotspots if data is skewed.
Hash partitioning: Hash the key, distribute by hash range. Even distribution. Loses range query ability.
Consistent hashing: Nodes on a ring. Keys map to nearest node. Minimizes redistribution when nodes change.
Rebalancing: When nodes join/leave, data must move. Can be automatic or manual. Automatic risks cascade failures under load.
Staff candidates explain WHY they chose a partitioning strategy, not just WHAT it is.