I was looking for online storage providers last night... they all charge an insane amount if you want to store large amounts of data (most were more than $100/month for 1TB of data... what??).  So I got to thinking, what about integrating Bitcoin into a distributed file storage system?

I believe I have heard of torrent-like file storage systems, where pieces of people's personal files (encrypted, of course) are stored across other users hard drives.  What about taking that sort of system a step further?

Files would be distributed across other users computers, but each user could be paid for volunteering their extra HDD space and bandwidth for the purpose.  They could set their own price (in Bitcoins) per GB of storage and per GB of bandwidth, and anyone needing to store files would automatically be given the best price for storage.  Sort of like a GPUMax for online storage.

Of course, you'd want each piece of the files to be backed up in more that one location, so that they are accessible regardless of whether some of the hosting computers are turned off or not.  Maybe people could pay extra for extra backups too.  2x copies would be standard, 3x copies could be paid for with 50% more on the price, 4x copies 100% more, etc.  And any host falling below a specified amount of uptime during a given time period would be booted from providing and/or not get paid.

The main reason for the project is this:  I think individual users are able to, and much more willing to, sell online storage space for much less than a hosting company would.  For example, I'd be willing to rent out 100GB of space with 500GB of bandwidth for $5/month.  Dropbox will do the same for $20/month.  Add another user like me, who is willing to rent out 100GB of space for $5/month, and you have your two copies, but it only costs the user $10/month instead of $20/month with dropbox.

Thoughts?  Would this be doable/practical/feasible/useful?

So, maybe something like Freenet's data storage scheme (maybe without the possibility of old data being removed), but with financial incentive to provide storage space and a free market system to ensure that using it stays cheap? Sounds like a great idea, and a great place to integrate bitcoin transactions.

I love the idea, I can sell out the TB's of hdd space I have on a vps.

This is an interesting idea.

The project is dead now but from what I can remember the fundamental economic model was broken. You had to pay upload and to download and the only ways to earn currency was to donate hard drive space or run indexing and directory services. They never could figure out a good pricing mechanism that actually worked.

It's been so long ago that I can't seem to find any reference to it and nobody I've asked can remember what it was called.

Found it:

Mojo Nation

It was extensively covered on Slashdot in the 2000-2002 timeframe.

When the company that was developing it started to run out of money they tried to take it commercial and turn it into a LAN backup product called Hive Cache.

One of the developers forked it as Mnet but without the micropayments. The source for Mnet is still available so someone who was so inclined might be able to revive the project and incorporate Bitcoin. The original Mojo Nation code was LGPL and there are still copies of it floating around too.

Given the leanings of the founder of that project it might even be possible to get the original creators on board.

The idea was good but it was about a decade too early...

https://tahoe-lafs.org/trac/tahoe-lafs

One of the developers (Zooko) is into bitcoin.  Perhaps starting a discussion about this on the tahoe-dev list would be well received.  There may be a way to easily integrate servers advertising a bitcoin address to encourage them to provide more space.

The incentives would need to be thought through carefully. If it was a simple matter of paying people to accept data then it would be attractive to scammers who would just throw your data into a bit bucket and take your Bitcoins.

If it is intended to be a distributed alternative to DropBox based on the Mojo model the people offering storage space would need an incentive to keep their nodes online and connected, which means an income stream proportional to the amount data they are holding and their node's past performance.

Block servers could all advertise a daily storage cost per block in an auction and the nodes seeking to upload data could send their blocks to the lowest bidder.

Each day (or whatever time interval is used) before a node issues payment for block storage it should query a percentage of random blocks from each block server to verify that the block server actually has the data. If it fails to deliver the requested data the block server would first not get paid and second that information should be fed to a reputation system.

If a user doesn't pay for storage it would be up to the block server to decide what to do. It could discard the data immediately, or it could retain it and demand back payments to release it, or most likely it would keep it until it was either paid for or until the block server ran out of space and unpaid blocks were overwritten by paying blocks.

So in the end the user wanting to store data tells his node the maximum amount he is willing to pay per day to store his data and how much redundancy/reliability he wants and the node automatically tries to find the most cost effective solution based on the block servers which are advertising free space and their reputations.

A straight swap still seems simpler. Want one of your blocks back? Gimme one of mine. Tell me which of yours you want and I'll reply telling you which of mine I want. Probably first wanting just a specific function of the block, only progressing to gimme the entire block if actually the whole block is needed/wanted not merely indication that the other party has enough of its bits to be able to answer the specific query.

"You want me to tell you what the bits at each of that list of 1024 bit-offets into the file are? Fine, I can do that if you undertake in return to tell me what the bits at each of this list of 1024 offsets into one of my files is".

If you are actually wanting cheap storage you buy, then a bunch of corner-store / mom-and-pop / kid in basement operations are probably not a cost-effective solution. A Wal-Mart approach is more likely to be able to get prices down and keep them down.

If you actually want the small private operations setup for reasons other than actual cost, then buy as much storage as the total of local and offsite storage you want, then trade storage with others who do the same.

I know that arguing not to use bitcoin to account these swaps might seem to imply my response is not bitcoin-related, but bear in mind I am looking to do offsite backups of bitcoin-related things such as bitcoin exchange servers and such. The relatedness / relevance to bitcoin lies not in method of accounting for the swaps of backup-storage-space but, rather, in that it is bitcoin users looking to backup bitcoin-wallets and bitcoin-websites and bitcoin-exchange servers and such.

It seems overly complicated and off-purpose to go looking to sell people storage space on my site to raise funds with which I can then go looking for other sellers to buy storage from on their site. It is simply I want to put N copies of X gigabytes offsite and am prepared to store X * N gigabytes similarly for others.

Of all the software suites I have looked at so far for accomplishing this kind of thing, so far GNUnet is looking like most appropriate. Though I suppose in principle I could also throw backups onto Freenet too if I chose to, what the heck, who knows, I might be able to find them there later if it turns out GNUnet fails me somehow. Maybe use GNUnet and Tahoe-LAFS and, just to be nasty to Freenet folk, throw copies onto Freenet too?

See that last part though, ouch, heck, what stops anyone from just leeching off of Freenet? Hmmm.
 
It might well be that the seemingly high prices the original poster is trying to avoid are not so high really at all when you really look into the nitty gritty of providing such a service.

It might also be that prices are driven crazy-high by involving the ignorant masses while not charging by the hour for help desk or other support services, thus having to do handholding of millions of idiots who do not want to pay by the hour for handholding thus basically force you into charging everyone for some high average number of hours of handholding in order to cover the biggest idiots who require the most-inordinate amount of handholding.

Offering the service only to the qualified I.T. departments of businesses maybe the price per terrabyte-month + price per terrabyte-mile could be held down below what those trying to attract the ignorant as their desired customers are forced to charge?

-MarkM-