It doesn’t necessarily have to be a datacenter. They’re just convenient for big companies spending huge amounts of money.
All a datacenter is is rows buildings filled with rows filled with racks filled with computers. Those computers could be distributed around the world in smaller buildings, or even sharing space with existing buildings that do other things. It’s not the fact that the data is stored in buildings filled with rows filled with racks that makes the data safely stored. In fact, that single location means that the data is more vulnerable than if it was stored in a more distributed way.
Having said that, keeping a lot of computers in a single building often does make it cheaper to run them because of easier maintenance and economies of scale. That means the data can be stored more cheaply, and it can be accessed more quickly.
But yeah, some datacenters is probably fine. I don’t think people were complaining much about Google’s various datacenters before they went wild on their AI push, and they had dozens of DCs spread around the world. What’s bad is that the current push is for many, many more DCs being built way too quickly in places that aren’t appropriate (because they can’t get enough power for example), for a purpose that people aren’t behind.
It doesn’t necessarily have to be a datacenter. They’re just convenient for big companies spending huge amounts of money.
All a datacenter is is rows buildings filled with rows filled with racks filled with computers. Those computers could be distributed around the world in smaller buildings, or even sharing space with existing buildings that do other things. It’s not the fact that the data is stored in buildings filled with rows filled with racks that makes the data safely stored. In fact, that single location means that the data is more vulnerable than if it was stored in a more distributed way.
Having said that, keeping a lot of computers in a single building often does make it cheaper to run them because of easier maintenance and economies of scale. That means the data can be stored more cheaply, and it can be accessed more quickly.
But yeah, some datacenters is probably fine. I don’t think people were complaining much about Google’s various datacenters before they went wild on their AI push, and they had dozens of DCs spread around the world. What’s bad is that the current push is for many, many more DCs being built way too quickly in places that aren’t appropriate (because they can’t get enough power for example), for a purpose that people aren’t behind.