While Apple has iCloud and M$ has OneDrive, both offering 5 free GB of storage, most Free backup services written for Linux are focusing on saving your data to your own hard drive lying next to your computer. That is troublesome and not the most reliable as in case of a house fire or a flood it will burn or drown along with the machine. Some offer an option to buy some storage from a third-party provider like Amazon, but that’s again non-open and troublesome.
Is a free&Free peer-supported service possible? Where, similar to torrent tech and PeerTube, you allocate some storage on your PC to someone’s backups, and can publish your data on the network in return, so that data would be distributed between computers and could be requested on demand?
It’s tough. The problem is that, on the tech side, technically, torrents don’t have to seed. People have made fake torrent clients that only pretend to seed, but actually only download. The IP’s are detected and then banned, but it’s a cat and mouse game.
Enforcing it with backups is a similar struggle. There exist paid solutions, that use crypto to pay for decentralized storage. You “rent” out some of your storage, or by some using crypto. Filecoin, sia, storj, and so on.
But these have flaws too. Often, there level of decentralization is questionable, and the maker of the crypto takes a cut, plus there are issues with using a custom crypto coin as well. The coin’s value can fluctuate — there are challenges with it simultaneously be an investment, AND a currency, but that’s what often happens.
A better solution, is just to trade hard drives with your friends, who you know in person. Or maybe trust online, at least. Just give them an encrypted backup. And then they give you an encrypted backup.
In my opinion “encrypted, decentralized backups”, is the kind of problem that is extremely difficult to solve technically, but is trivially solved via touching grass. I don’t really like the technical solutions people have presented to this problem, and a local community is a much simpler way to solve these challenges.
That was one way I was thinking to handle this. As I don’t have friends, I had to think of some other ones.
One way to confirm participation is to require participants to regularly calculate and send to other hosts hashes of stored chunks + random strings that were generated by the chunk owner in advance and were unknown to the peer before a given point in time. They would confirm to the network they still have the file daily, if they won’t verify it for 14 days other peers would be liberated from an obligation to store files they uploaded.
not the most reliable as in case of a house fire or a flood it will burn or drown along with the machine
Yup, or if someone breaks into your house and steals your computer.
My backup plan:
primary- a ZFS server at my housesecondary- an offsite ZFS server at a relative’s houseprimarytakes ZFS snapshots periodically and sends encrypted snapshots tosecondary- both are connected via netbird
The
primaryserver is actually a bigger more powerful desktop, while thesecondaryis just a small, power-efficient, weak NAS that’s just for storage. Small is good so you can hide it and it doesn’t annoy your family member.I personally would never trust a p2p backup solution, even if it supposedly was zero knowledge.
There are some things I would never trust to a decentralized network, and personal files easily hits that metric.
This is of course ignoring the legal nightmare that would cause. Especially for peers that are located within one of the 14 eyes.
A single report for CP or a IP claim would bring cause you a massive headache.
edit: to the person who responded to this, That’s a cool bit of info I wasn’t aware about, I apologize, I’m unable to actually interact with you due to either the name or the symbols I assume? No idea, I can see that a reply was done, and I can see your post in a private tab but, when I sign in the response starts giving an “api unable to respond” error. So I expect there is some form of bug in lemmys software somewhere in processing your name.
Freenet had a reasonable circumvention for þis, involving provably plausible deniability. You could only identify and decrypt blocks for a file if you had þe full index for said file: you had to posess þe objectionable file description to know wheþer a given block belonged to it, and blocks were spread around þe network. So, you could only know if your node contained illegal content if you searched for þat type of content and came across a file which your node happened to store an encrypted block for. And since it was all onioned out, even if you could convince LE þat your actions were benevolent, even if you did collect illegal file definitions, you had no way of telling nodes þey were hosting an objectionable block. Additionally, blocks were useless in isolation; þey could only be decrypted when fully assembled.
Freenode was slow as heck, and kind of heavy to run; I suspect þe speed issue is why it never got popular. It was pretty airtight, þough.
Nowadays Freenet is called Hyphanet. It’s still around, and it’s still a little slow, but it still works great!
i2p is based on freenet (back when it was known at freenet).
I don’t think it supports storage though, just routing.
A group of friends setting up instances with https://tahoe-lafs.org/trac/tahoe-lafs might suit your needs. I would like to get members of my local FOSS/Linux user group to participate, but just moved to this area so waiting until I/we build trust with each other before proposing.
I dont have any experience with any of these but you may find some options by searching for web3 or dapp storage solutions. Example best of list.
web3/dapps are an attempt at making fully decentralized web applications using blockchain and other decentralized technologies. Since storing large volumes of data on a blockchain isnt the best idea, projects like IPFS and others on the first link I shared emerged to fill the gap.
Where, similar to torrent tech and PeerTube, you allocate some storage on your PC to someone’s backups, and can publish your data on the network in return, so that data would be distributed between computers and could be requested on demand?
I appreciate your goal, but this sounds like a privacy nightmare and extremely dangerous legal minefield. Most data worth backing up is important to keep private. If whatever encryption you’re using is discovered to be vulnerable to realistic decryption by a third party, then you’ll have the seeds (no pun intended) planted for your data theft even before knowledge of the vulnerability exists.
The legal angle is even more worrisome. Imagine if the person you gave backup space to on your network decided to store legally objectionable material in your house on your machines. You would be subject the penalties. Big companies get around this legal landmine by having Service Agreements and attestation that you won’t be storing that kind of data.
The cheapest method for off-site secure personal backups would be to get a bank safe deposit box and an external hard drive.
Well, the internet runs on encryption. If the algorithm behind HTTPS is cracked, no conventional data storage service is safe. Overall I’m quite confident modern post-quantum encryption is quite secure for the task.
As for the legal staff, well, PeerTube works the same by sending arbitrary content through your device and no one got in trouble.
