Part 1 of 10 · 1 min read
Why this page comes first in the toolbox
In this module you will
- Describe, in order, the changes a system makes as users grow from one to millions, and the problem each change solves.
- Trace one request from a browser to a server and back, including the address lookup.
- Tell vertical scaling from horizontal scaling, and say why large systems prefer the second.
- Explain what a load balancer, a replica, a cache, a content delivery network and a queue each fix.
- Explain why servers should keep no user state, and what "stateless" buys.
- Say when a database must be split across machines (sharding), and name its three common problems.
- Use the ladder in the interview as a map of "what would I add next, and why".
Every later level takes one rung of this ladder and goes deep. Level 1 covers load balancers and content delivery. Level 2 covers databases. Level 3 covers caches. Level 4 covers replication and sharding. Level 6 covers queues. Level 9 covers logs and metrics. This page shows the whole ladder once, so that each later piece has a place.
The numbers on this page are made up to show when each step is needed. Real limits depend on the hardware and the work.