Vertical scaling

Vertical scaling increases CPU, RAM, or disk on a single node. It wins early for simplicity: no sharding logic, easier transactions, and lower operational surface. Limits appear at hardware ceilings, cost per core, single-point-of-failure during upgrades, and inability to exceed one machine's I/O.

Combine vertical scale with read replicas or caching before jumping to sharding. Document when you will outgrow the largest instance class.

On interviews: argue when vertical scale is the right MVP choice, name hard limits, and describe the trigger to go horizontal.

Common pitfalls: delaying shard design until emergency; assuming cloud instances scale infinitely; downtime during vertical resize on some platforms.

Checklist:

• State current and max instance headroom.
• Compare cost versus horizontal alternative.
• Plan failover during vertical upgrades.
• Define metric threshold to scale out.