Best practices for S3/Cloud Storage On a Budget

Questions
1

Hello everyone. I apologize since I’m pretty new in handling discussions of this topic with people who are more knowledgeable.

In summary, my intention with this article is twofold:

  1. Explain the ideas I’ve had to have decent privacy and usability of cloud storage
  2. Keep things within a budget without compromising much on the first.

So first, the reason why I need S3 Storage and Cloud Storage is that I have twofold needs: A way to keep my public-facing data agnostic to a system, and my private data secure and available on-the-go.

Unfortunately, this discards any sort of self-owned cloud-attached storage. While long-term backups are done offline, I do need a part of my data available for freelance work. I also use Linux, so most providers don’t fit my purpose.

Ideal options would be:

Small/cheap Cloud Storage, Cheap/Free S3 storage

Public work does not exceed 10GB so Free-Tier S3 storage is viable. My private projects need about 50GB, but long-term storage for about 200GB would be nice.

Robust S3 Storage with E2EE client

I could also go the more advanced route and keep everything on an S3 Storage, while all interactions of the private side use E2EE. It would also enable me to expand private projects (Ente, for example)

For this reason, I’ve considered a few candidates.

Proton Drive

I already use several tools on the suite, but with the Linux client being ever out of reach, I could navigate a few months with no cloud storage. It does look like the most likely alternative.

MEGA

I don’t like it, but it is cheap. With a second layer of E2EE some of the glaring issues can be mitigated. Since it also features S3-Compatible access, it looks like the best “Bang for the buck”, but so many people have told me to stay away from it, even my own instinct, it is more of a “bemusement” option than a serious consideration.

Filen

I’ve seen lots of mixed opinions regarding this, but the affordability and feature-set does seem tempting for a test drive.

If you have a similar situation or any insight you’d like to provide, I’d be happy to hear from you.

2

Proton Drive for Linux is being developed. Depending on what you need the Proton Drive CLI could also be good enough. People have also reported using rclone is somewhat stable bow but I’m not sure about that.

3

I did see the CLI release, but I didn’t really think much of it since I don’t use Proton Drive. However, since the Linux client has been a priority in posts as far back as 2024 so I’m not really sure their current “Priority for 2026” seems likely. But knowing my luck, it’ll probably release in 2 weeks and I’ll come back to this post with egg on my face.

4

I don’t really understand this question. What part of S3 is appealing to you?

Are you hosting apps which need S3 compatibility?

Do you need the high availability that Amazon S3 provides so that it’s never offline? (Therefore self-hosting storage would probably not be an option)

5

Yes, as mentioned, I have public-facing data that needs to stay system-agnostic. That is to say I don’t want to be locked in to a single provider in case it needs to be migrated, be it from the endpoint it is accessed from, or the system itself I use.

I also mentioned I don’t use more than 10GB for the public-facing data and systems, so it’s a low concern. If my needs change, it probably won’t be for a while.

Though it would probably be more economical and give me availability to expand private projects or ones that don’t require even 80% uptime.

Local storage is not an option. While I don’t need “Three nines” of availability, local infrastructure is not within my management capabilities due to a number of factors, namely the overhead cost to mitigate outages (Such as rolling blackouts).

6

I’d check Oracle always free tier. You ruled out cloud-attached storage, but at the free tier the maintenance overhead might be worth reconsidering, depending on what made you rule it out.

I don’t think consumer clouds like Proton or MEGA beat Backblaze B2 on price, unless you’re already paying for the Proton ecosystem anyway. S3-compatible B2 paired with restic/kopia/vykar gets encryption without needing a provider that offers E2EE natively. Though, this only solves backup, not live sync/on-the-go access. The latter can be achieved with rclone.

For the “on-the-go with own encryption” piece, another combo worth testing: Google Drive or other consumer clouds with linux support + Cryptomator or rclone crypt on top.

7

I don’t think you need to say “as mentioned” twice while including details you had not mentioned.

MEGA’s Pro plans (up to 20TB) are relatively price competitive, but I have never signed up with them. It’s really only a great deal if you need 3TB–20TB though, otherwise they’re basically the same as anyone else. If you went with them I would just ignore their regular storage clients and store your private data in Mega S4 with rclone crypt E2EE. Or you could do the same with Backblaze, Hetzner, Scaleway, DigitalOcean, Cloudflare, Google, Amazon, etc.

Obviously would not recommend the big tech options here on this forum though :slight_smile:

It pretty much comes down to pricing and features. Backblaze B2 may be 1/4th the price of Amazon S3, but S3 will store your data in multiple datacenters, whereas with B2 you’d have to send your data to two B2 regions yourself (thus making it only 1/2 the price for example) if you wanted that redundancy.

If you were self-hosting, you could just run Garage on 3+ computers for redundancy and availability.

8

Mega’s own tools have had some pretty noteworthy security issues and limitations which is why they typically aren’t recommended. But personally yeah I’d probably still use Mega S4 with separate encryption like rclone crypt if I needed to.

9

S3 services don’t provide a client, they provide an API. It’s up to you to shape up your data to your liking before storing it in S3.

Not S3, no programmable client, unusable in development context. Same for the rest of them.

10

forum discussions are funny