CDN — Part 2: Architecture, Caching & Edge Compute

What is this?

Part 1 sold you on the idea: put copies of your content in corner shops near every user. This part is about running the chain. A single coffee shop is easy. Hundreds of them, all stocking the right beans, throwing out what's stale, restocking from the warehouse, and — increasingly — doing real work behind the counter instead of just handing over pre-made cups? That's the engineering.

So we go one level deeper. How does the cache hierarchy actually decide what to keep when space runs out? How do you invalidate content across the planet without a stampede? And the big shift of the last decade: how did CDNs evolve from storing your files to running your code at the edge?

The one-line idea: a modern CDN is no longer a passive shelf of cached files — it's an active, programmable layer at the edge that decides what to cache, how to keep it fresh, and increasingly runs your logic right where the user is.

We'll keep the supply-chain picture from Part 1: edge = corner shop, mid-tier = regional warehouse, origin shield = the single loading dock that protects the roastery, origin = the roastery itself.