I interned for Google’s Cloud Storage team in Sydney a few years ago. Google is building over 100 new data centres every year, and the annual numbers are growing. The storage capacity is simply being built much faster than people are filling it up. Also, the existing data centres get updates to have even more storage capacity when better hard drives are invented.
Accept that a single machine can only store a limited amount of data. So, a common strategy to store more than one machine can handle is to… use more than one machine!
So, instead of storing all the data in a single machine, the data is spread across multiple machines. And, with engineering magic, this massive group of machines behave as if they were a single (and massive) filesystem.
Moreover, those companies not only provide an “unlimited” storage for their users but they also provide a reliable storage! Imagine that they were storing all your data in a single computer, and poof! The computer explodes. Are you going to lose your data? No! Because engineers thought about this and they decided that your data will not be stored in single machine but in several machines! So, even if a machine explodes, you data is still intact in another machine.
The more content you upload, the more they either:
– know you better and are able to direct ads tailored for you to click and buy. This option is a bit expensive because maybe you just won’t buy stuff but you’re still using their “unlimited” storage. Think Google Photos: they got to a point where new images weren’t as helpful as before, so they capped it.
– or you attract more people to see your content, and that creates sale opportunities not just for you, but to X amount of people who are following your content. YouTube does this. But it has recently been working with limits to storage, I believe it’s something like “if your video doesn’t have enough view, we’ll delete it”
Platforms will shuffle through strategies to achieve certain goals they have. Imagine you need a bunch of images to train AI, could then give incentives for users to upload all their pictures in high quality so you take pieces of it for recaptchas and then use people answering “I’m not a robot” to train your AI. Don’t need anymore? Say storage is now counting on your images.
Google some pictures of Google’s data centres and headquarters.
There are several of them, and “very large” doesn’t even begin to describe just how large they are. I’m talking about fully enclosed mega structures that make airports look like corner stores.
Most of them have roads and vehicles inside to get from one side to another!
Latest Answers