Cloud Region

When a company rents storage space or computing power over the internet instead of buying its own computers, this is called the cloud. These rented computers are housed in large halls full of machines, known as data centers. A cloud region is a group of such halls in a particular area of the world, for example around Frankfurt am Main. Anyone booking something with a provider must always specify which region it should run in. This choice determines in which country the data physically resides. It also determines how long a signal takes to travel from the user to the computer.

Why location determines both law and speed

Data is subject to the laws of the country in which it is stored. For European companies, this is a central issue because the General Data Protection Regulation sets strict rules. A hospital or a bank is often only allowed to store patient or account data within the EU. That is why the major providers offer their own regions in Frankfurt, Paris, Zurich, or Stockholm.

The second reason is speed. Even at the speed of light, a signal takes time to travel long distances. From Berlin to Frankfurt and back takes only a few milliseconds. To California and back can quickly take 150 milliseconds or more. This delay is called latency. In an online shop it is barely noticeable, but in a video game or a video conference it is very noticeable indeed.

The third reason is fault tolerance. An earthquake, a power outage, or a fire always affects only one location. Anyone who distributes their services across two regions that are far apart remains reachable in the event of a disruption. Major providers explicitly recommend this, because even they regularly experience outages in individual regions.

Regions, zones, and the networks in between

A region usually consists of several independent locations. These parts are called availability zones. Each zone has its own power supply, its own cooling, and its own internet connection. They are located far enough apart that a disaster will not affect both, but close enough for very fast connections. Typical distances range from a few kilometers to a few dozen kilometers.

You can imagine this like a supermarket chain in a city. The region is the city, the zones are individual branches. If one branch is out of service due to water damage, customers shop at the next one. In exactly the same way, the software automatically shifts operations to another zone.

Between the regions run the providers' own fiber-optic lines. Backup copies are mirrored to other countries over these connections. It is important to distinguish this from a so-called edge location: these are small intermediate storage points that only hold copies of frequently accessed content. They cannot compute and store like a full region. Another common misconception is that every region offers all services. New offerings almost always launch first in the USA and only reach Europe later.

Where cloud regions come up in the news and everyday life

Business news often reports on billion-dollar investments in new regions. Microsoft, Amazon, and Google are continuously building locations in Europe, India, or Saudi Arabia. Such reports are also politically interesting, because data centers require a great deal of electricity and water. In Germany, this has led to disputes over the grid connections of new facilities.

Anyone who sets up their own account with a cloud provider will immediately see the regions. They appear as a selection list with names like eu-central-1 or westeurope in the interface. Prices also differ noticeably by region, often by ten to thirty percent.

For AI, this topic has gained additional weight. Specialized graphics chips for training large models are scarce and available in only a few regions. Companies must therefore book their computing time wherever these chips are available. At the same time, European customers demand that their requests to a chatbot not leave an EU region. Providers now explicitly advertise such guarantees.