Database scaling handles increased data volume, traffic, and availability requirements.

Vertical Scaling

Make one database server stronger:

• More CPU.
• More RAM.
• Faster disk.

Simple but limited.

Horizontal Scaling

Add more machines.

More complex because data must be replicated, partitioned, or sharded.

Read Replicas

Primary handles writes; replicas serve reads.

App writes -> Primary
App reads  -> Replica

Tradeoff: replicas may lag behind primary.

Partitioning

Split a large table into smaller parts.

Examples:

• By date.
• By region.
• By tenant.

Sharding

Distribute data across multiple database servers.

users 1-1M    -> shard A
users 1M-2M   -> shard B
users 2M-3M   -> shard C

Sharding improves capacity but complicates joins, transactions, rebalancing, and operations.

Caching

Use Redis or application caches for hot reads. Cache invalidation becomes the hard part.

Interview Notes

• Vertical scaling is easier but has limits.
• Read replicas scale reads, not writes.
• Sharding scales writes/storage but adds complexity.
• Partitioning can improve manageability and query performance.
• Always measure bottlenecks before scaling.