Regions, zones and edges: the geography of the cloud

Regions, Availability Zones and edge locations sound interchangeable until you need them to mean different things. They are three layers, and knowing which is which is how you keep an app online.

AWS runs in a lot of places, and the words for those places sound interchangeable until you need them to mean different things. Regions, Availability Zones and edge locations are three layers, biggest to smallest, and knowing which is which decides how you keep an app online.

A Region is a place on the map

A Region is a geographic area, think Mumbai, Frankfurt, or North Virginia. Your data lives in the Region you pick and stays there until you move it, which matters when a law says customer data has to stay in the country. Choosing a Region comes down to four things: any compliance rules about where data can sit, how close it is to your users (closer means lower latency), whether the service you want even exists there yet, and price, because the same service can cost different amounts in different Regions.

Availability Zones are separate buildings in that place

Each Region is split into Availability Zones. An AZ is one or more data centres with their own power, cooling and networking, far enough apart that a fire or flood in one will not take the others, but linked by fast fibre so they still feel close. The whole point is failure isolation.

So the single most useful habit in cloud architecture is a boring one: run your app in at least two AZs. If one goes dark, the other keeps serving and your users never notice. A single AZ is a single point of failure wearing a cloud costume.

Edge locations sit closest to your users

Then there are edge locations, and there are many more of them than Regions. They are small sites whose job is to cache content near people. Put CloudFront (AWS's CDN) in front of your site and your images and files get copied to the edge location nearest each visitor, so someone in Tokyo is not waiting on a server in Virginia.

Region is the city. Availability Zones are separate buildings in it. Edge locations are the corner shops on every street.

That is the whole hierarchy. Pick a Region for law, latency and cost. Spread across AZs so one data centre failing is a non-event. Push content to the edge so it loads fast wherever your users happen to be.