What actually happens when a CDN caches your page

I'd read the diagrams a dozen times: browser hits edge, edge checks cache, cache miss goes to origin, response gets stored, next request is a hit. It made sense on paper. It didn't feel real until I watched it happen against a page I control.

The first request after creating the resource came back with a cache status of MISS — the edge node had nothing stored yet, so it pulled from the origin VM, served it, and kept a copy. Every request after that, from anywhere near that edge node, came back HIT without touching origin at all.

The part that isn't obvious from a diagram: a HIT and a MISS at the CDN layer are invisible to the end user in every way except one — origin load. The response looks identical. The only reason to care is that your origin now handles a fraction of the traffic it used to, which is the entire point.

The other thing worth internalizing: cache state is edge-node-specific, not global. Whether the visitor next to me sees a HIT or a MISS depends on whether their nearest edge node already has a copy, not on whether the object exists anywhere at all in the CDN's network.

curl -I https://cdn.miska.blog/
> HTTP/2 200
> cache: HIT

That single header line is doing more explanatory work than any architecture diagram I'd looked at before actually standing one of these up myself.