Forty lessons on how a system behaves when traffic arrives. Start with a single page load, then add one pressure at a time — a cache, a shard, a replica, a queue, a timeout — and keep your eye on P50, P99, throughput and CPU.
Nine modules, forty lessons, one question: what happens to a number when the architecture changes? The course starts with the path a request takes — name lookup, connection, request, response — and ends with four complete designs (short links, a news feed, chat, and the trade-offs of high-level vs low-level design). Caching, sharding, replication, queues, consensus, failure and recovery, capacity planning and cost per request sit in between, each with its own lesson and its own figures.
Every lesson opens with the pressure it is about to apply — “users say the shop feels slow”, “one shard takes twice the traffic” — and then walks the same scene step by step. You see which part of the budget moved: the extra round trip, the queue that grows when a consumer falls behind, the P99 that doubles when a key goes cold. The point is not to memorise a diagram; it is to be able to say what a design costs and when the cost stops being worth it.
Backend and full-stack engineers who can already ship a service and want to reason about latency and capacity before the incident review, people preparing for system design interviews, and anyone told to “just add a cache” without being shown what that does to the tail. You need to know what a request is and to have read a metrics dashboard; you do not need a cloud certificate or a particular framework.
The first two lessons are free, no account needed. The full forty lessons are a one-time $4.99 on the account you sign in with, and progress is stored per account. This course is bought on its own — it is not bundled with the protocol explorer, Learn Bitcoin or Learn Arduino, and a Pro pass does not include it. Refunds within fourteen days by email.