Website scalability

Business basics2min read

Website scalability is a website's ability to handle more visitors, requests or data without unacceptable slowdowns or failures. It depends on the whole system, from application servers to databases, and on how much extra capacity costs.

How it works

A fast website is not necessarily scalable. It might load quickly for ten visitors but struggle when thousands arrive together.

Scalability depends on the bottleneck. Adding application servers will not fix a database that cannot keep up. Caching repeated requests, improving database queries and moving slow tasks into background jobs can increase capacity before you buy more servers.

Handling 600 requests per second

Here is an illustrative capacity model. Each application server costs $50 per month and handles 200 requests per second at your target response time. Capacity rises in direct proportion to server count, with no database or external-service bottleneck.

Application serversModeled capacityMonthly server cost
1200 requests/second$50
3600 requests/second$150
4800 requests/second$200

Capacity equals server count × 200; cost equals server count × $50.

At demand of 600 requests per second, three servers leave no spare capacity. Four provide 200 requests per second of headroom. If one becomes unavailable, the other three still have enough modeled capacity.

These costs exclude databases, bandwidth and other services. For a startup, the useful claim is tested capacity at a stated cost, not simply "built to scale." Validate the whole system before promising launch-day performance.

Questions people ask

How to make a web app scalable?

Start with a load test and measure response times, errors and resource use as traffic rises. Fix the first bottleneck, then test again. If you add application servers, keep shared data and user sessions accessible across them.

What are the two types of scaling?

Vertical scaling gives an existing server more resources, such as CPU or memory. Horizontal scaling adds servers and distributes traffic between them. Vertical scaling has a machine-size limit; horizontal scaling requires software that works across multiple machines.