Module 04 · System DesignThe parts, properly

Why Data Is the Hard Part (the five building blocks)

Data Kya Hai (Why It Matters) · Open board

Numbered last because it was drawn last, but it explains what every other board in the track is for: whichever wall you hit next, this is the board that named it first. What a Data System Promises draws the identical five-block map from a different angle - the same shape, found twice.

Before you start. You need only arithmetic - multiplying, dividing, and reading a percentage as “so many out of a hundred.” Nothing here leans on another lesson.

The idea in 30 seconds

Data-intensive means the bottleneck is the data, not the arithmetic - test it by asking whether doubling every CPU makes the problem vanish.
Every app runs store, read, generate - and the biggest slice of what gets stored is what people tapped, not what they bought.
A single machine hits four walls, and the one that finishes it first is never the processor - it’s the disk filling on schedule, or the one copy going down.
A machine can go from fine to unusable well before it’s full, because wait time grows faster than the traffic causing it.
Replication fixes “one machine can’t survive” and buys a new problem: which copy is right. Partitioning buys a slow worst case, because a query across every shard waits on the slowest.
Exactly one copy can’t be rebuilt, the record itself; every other store - cache, index, stream, batch job - is disposable, fed from the record’s own log.

Read the story · Skip to the numbers

Continue reading with a plan

The first three lessons of every track are free. This lesson is included with a plan, which unlocks every lesson, board, video, lab and paper for one price.

View plans →Already a member? Sign in

Review System Design →

Connects to

Saved in this browser, and to your account when you sign in.

← Thirty stalls on one order (availability, the nines, and the tail at scale)The Right Book for the Job (databases, deeply) →