It's Not Just "The Edge"
Think of CloudFront like a postal system. When you order something online, the package doesn't travel directly from the factory to your door every time — it moves through regional distribution centers, local sorting facilities, and finally a delivery truck in your neighborhood. CloudFront works the same way: requests from users travel inward through progressively larger caching layers before ever reaching your actual servers.
When most people think of CloudFront, they think of "Edge Locations" (Points of Presence, or POPs) — the thousands of servers physically close to end users. But CloudFront is actually a multi-tiered architecture designed to absorb traffic and protect your backend (called the Origin) from being overwhelmed.
The three tiers are:
- Edge Locations (POPs): The outermost layer. Thousands of servers distributed globally, physically close to users. This is where cache hits are served with the lowest possible latency.
- Regional Edge Caches (RECs): Larger caches that sit between the POPs and the Origin. When a POP misses its local cache, it checks the REC before going all the way to the Origin. RECs serve multiple POPs in a geographic region.
- Origin Shield: An optional, centralized "super cache" layer you explicitly enable. It sits between all RECs and your Origin, acting as a single point of contact for cache misses.
Understanding these layers is the difference between a 90% cache hit ratio and a 99% cache hit ratio — and the difference between your Origin surviving a traffic spike or collapsing under it.