Compare vertical and horizontal scaling. When would you choose each?

Vertical scaling (scale up) means using a bigger machine. Horizontal scaling (scale out) means adding more machines.

Vertical scaling gives one machine more CPU, more RAM or faster disks.

• Pros: it needs no code changes and keeps everything simple.
• Cons: it has a hard ceiling, because machines only get so big. It gets expensive at the top end. The machine is still a single point of failure: if it dies, everything stops.

Horizontal scaling puts more machines behind a load balancer. The load balancer spreads requests across them.

• Pros: it has no practical ceiling. It also gives redundancy, because other machines keep working if one fails.
• Cons: the application must be stateless, or keep shared state outside the machines. You also take on distributed problems, such as data partitioning and consistency.

When to choose each:

• Scale up first when it is cheap, especially for databases.
• Design the application tiers to be stateless, so they can scale out.

Stateless services scale horizontally easily. Stateful stores, such as databases, are where horizontal scaling gets hard.