How to work through this course and choose your path

This course contains twenty-five worked designs. Reading them is worth far less than attempting them, and the difference is large enough to change your result.

It is also about sixty hours long, which is a lot to spend if half of it is not relevant to the interviews you are actually taking. This lesson answers both problems. First, how to study a section so it actually changes what you can do in the room. Second, which sections to study: everyone does Parts I and II; after that, choose.

Part I in order, then attempt-first for everything else

Sections 1–5 are foundations and are meant to be read straight through. Section 3 (estimation) and Section 5 (building blocks) are the two you will return to constantly — Section 5 in particular is a reference, not a narrative, and every later section links back to it rather than re-explaining quorums for the twelfth time.

From Section 6 onward, every section opens with a problem statement and an instruction to stop reading. Follow it.

The protocol

1. Read only the section title and the problem statement in its first lesson.
2. Set a 45-minute timer. Use paper or a whiteboard, not an editor.
3. Work the four steps from Section 4: scope, estimate, design, deep-dive.
4. Say your reasoning out loud, to an empty room if necessary. The verbal channel is the one being assessed and it is trained separately from the thinking.
5. Then read the section, and mark every place your design differed.

The marks are the value. A difference is one of three things: you missed a requirement, you did not know a building block, or you made a defensible different choice. The third is a good outcome and you should be able to tell it apart from the first two.

Why passive reading fails here

Reading produces recognition — the feeling that you would have thought of it. Under time pressure, with someone watching, recognition does not convert into production. The distinctive failure is the candidate who has read about consistent hashing, nods when it is mentioned, and cannot construct the argument for why hash(key) % N breaks (Section 7, The problem it solves) when they need it.

Pacing

The course is about sixty hours of reading and around eighty with practice. Four to six sections a week is a sustainable pace and puts a full pass at six to eight weeks. If you have less time, the rest of this lesson tells you what to cut.

Do at least three timed mock designs with a partner before a real interview. Practising without an interviewer, in Section 31, has the rubric to hand them.

The fixed core

Part I (Sections 1–5) is the framework, the arithmetic, and the vocabulary. Nothing later makes sense without it.

Part II (Sections 6–11) is six small systems, each carrying one hard idea: rate limiting algorithms, consistent hashing, storage engines and quorums, distributed identifier generation, the read-heavy key-value system, and a large pipeline. These ideas reappear in every case study in Parts III–V. Skipping Part II to get to the "real" systems is the most common mistake learners make with this course, and it costs them the deep-dive axis.

That core is Sections 1–11: roughly twenty-two hours of reading.

Then choose by role

Everyone finishes with Section 31 (The Learning Continues), which extracts the recurring patterns across all twenty-five designs and gives you a self-assessment to find your own gaps.

If you have one week

Sections 3, 4, 5, then 10 and 13 as worked examples, then 31. That is roughly nine hours and covers the framework, the numbers, the vocabulary, the most-asked warm-up, and the most-asked hard question.

If you have one evening

Section 4 (the framework) and Section 3 (estimation). Being able to run the four steps and size a system in three minutes will not make you look experienced, but it will stop you freezing, which is the more common failure.