Databases for System Designs
Database scaling in system design is about growing storage and throughput beyond a single server. Common techniques are replication for read capacity and availability, partitioning or sharding to spread data, and indexing. Choosing a store means matching access patterns to relational, key-value, document, wide-column or graph databases.
Stage: Data at Scale · Part of the System Design roadmap · Last reviewed 19 November 2026
What to learn in Databases
- Set up leader-follower replication and describe replication lag
- Choose a shard key and predict hot spots
- Explain consistent hashing and why it limits data movement
- Compare SQL and NoSQL stores for a stated workload
- Decide when denormalisation is acceptable
- Describe the cost of cross-shard queries and joins
Databases interview questions
How do you choose a good shard key?
Pick a key with high cardinality that spreads load evenly and matches the most frequent queries, so most requests hit one shard. Avoid keys like timestamps or country that concentrate traffic on a single partition.
When would you pick a NoSQL database over a relational one?
Consider it when the access pattern is simple, such as key lookups, when the schema varies, or when you need horizontal scale beyond one node. Choose relational if you need joins, multi-row transactions and strong integrity.
Where Databases fits in the System Design roadmap
Databases belongs to the Data at Scale stage (advanced level), which takes about 32 hours. Store and move data.
Other units in this stage
Back to the full System Design roadmap