On Wed, 12 Mar 2025, RAGINI wrote:
3. Why use S3?
Here is the verbatim copy of the content from 'djbon2112' response It's shared storage over HTTP(S) basically. There's a couple reasons this is beneficial in webscale applications: It's shared. Unlike simple "files and folders on a filesystem", it can be accessed by multiple systems at once without using storage-specific protocols like NFS. It's dynamic. You just put data into it. No worrying about a volume filling up or anything like that. It's all abstracted away. For commercial object storage providers they bill you on what you actually use, rather than the size of a disk that you'd probably want to keep under 80% utilized at all times. It enables more client-side focused interfaces. Imagine an app on a phone. You have your database backend, your API servers, and then you store all your binary data (e.g. images, etc.) in Object storage. Under a "traditional" storage scheme, you'd have to mount your shared storage for that binary data on all of your API servers, and then serve it along with the content. In effect, you're proxying all requests for that binary data through your app servers, which would amount to a large percentage of the data transfer done there. With object storage, you just send the client a link to the object storage bucket and it can fetch the images itself. This also helps massively with scale, since requesting large files can tie up app servers and limit their request rates. It's not a solution for every problem, like most things it has its uses and its anti-uses. But a lot of the hype is around the things it enables in terms of scalable datastorage with a client focus. URL https://www.reddit.com/r/selfhosted/comments/10tzhmw/eli5_why_the_hype_on_s3object_storage/ warm regards Saifi.
