This blog entry from 2006[1] is only 13 years old, but feels like it should be older. Back then, putting that much data for a publicly accessible site was considered pretty insane.
[1] https://don.blogs.smugmug.com/2006/08/12/amazon-s3-the-holy-...
S3 was invented in 2006.
What was S3 used for if not for hosting website data?
Because looking back at that, it feels odd that it's only 13 years ago, and how much the industry has changed in that time.
Back then S3 was this weird thing that Amazon came out with. A file hosting service that didn't let you do anything except store or fetch files.
Sure, it's low cost... but I can't process them without downloading them again and paying for the data access (twice!). Why the hell would I want that?
SmugMug's use case was the first I saw that actually made sense.
Back then, if you weren't able to afford buying a server and putting it in a colo facility, you'd go rent a server on a monthly/yearly basis from the likes of ThePlanet, and host those files yourself on your server. If you needed redundancy, you'd have to figure out syncing between the servers yourself.
Now? Spin up an AWS instance or trigger a function, serve a link to the file in S3 or via a cloudfront distribution. Even if you need to process the data before serving it - stream it straight out of S3, through your application and to the user.