Distributed cloud storage... this looks interesting

sk66

sk66

Suspended / Banned
Messages
9,557
Name
Steven
Edit My Images
Yes
Don’t QNAP do something similar with the Nas enclosures and sharing storage across multiple devices/location?

It certainly looks like an interesting concept and using P2P should help with downloads
 
OK, full declaration, I have a fair insight into the technologies involved as part of my day job.

My impression is that what they are talking about is possible with current technology, but I fail to see what their USP is. Plus, you are welcome to spunk up the one off cost but I honestly can't see the viability of a one off non-subscription system, they are creating the digital equivalent of a Ponzi scheme, that's a hard truth.

You would be better off buying into AWS or the like.
 
Actually, no. It's striped across machines, decryption from a single point is not possible because, well, it's just not there.

If you know how CD are encoded, it's like that but instead of the data being 180 degrees apart on the CD it is in two different locations.
 
Shared network work fine, as does distributed file sharing, TOR is built around that. But that isn't the same as keeping YOUR files for nothing.

Let me explain. Distributed file sharing works because 100 people have copies of Debbie does Dallas on their PC, so if you want a copy you can suck a bit from each of those, and you know what? As an added bonus you are the 101th.

What incentive is there for anyone to host MY storage? none. It will work as long as enough people join with enough space to keep servicing the need, that's why I call it a ponzi scheme. It is possible that I've not read it properly, but that's how it seemed to me. Oh yeah, plus there is far too much marketing woo for me to feel comfortable. As I say, I understand the tech, and I see no mention of it.
 
Last edited:
One of the throwaway lines in the campaign is that by adding a 2Tb drive to your home unit, you gain 1Tb of space, which goes along with the idea of striping data across multiple units. However, when a 'standard' home user buys a 4Tb drive, thinking they're going to gain 4Tb of usable storage, they're going to be complaining that they only see a 2Tb gain.
 
With redundant data encoding, you aren't mirroring, there is only ever a single "copy" of your data, but the algorithm exists to recreate the data in its entirety if part of it disappears. The trouble is, that isn't a zero sum game, there is a substantial calculation overhead and time lapse before it can be recovered. With multiple domestic data stores, I don't see that coming easy. The more datastores involved the more complicated the algorithm gets. For something like this I reckon there may be two copies of your data. You might well have a single intact copy of your data locally, and the remaining 50% is the second copy distributed. Only, it will depend on you not using the whole of your local 50% because the remote storage will take more than 50%.. While that's a pretty safe assumption, it is still a ponzi scheme.
 
Sounds too good to be true, but if this is so wonderful and works with current technology, why hasn't it already taken over from current cloud storage?

If it does indeed work as advertised, then it surely will soon. I'll wait till then thanks (and no, I haven't a clue how it works).
 
So they sell a few in the first uptake as it all new and shiny, then sales dry up and they will very fast, and they simply walk away or cannot afford to maintain the software and hardware, You are left with problems and a system that will die, and they probably made some cash. Monthly subscriptions are where business models work, it’s a huge asset and a constant revenue stream, however is suspect the hardware cost for a startup would mean a large monthly fee.
 
Too many questions...

Headline features are "Privacy by design, forever-free, lightning-fast." But it's far from free, and I'm not really that bothered about privacy or speed.

First and foremost, we need long-term security at reasonable cost and we've already got that from numerous and reliable cloud providers, tried and tested. I don't want yet another box that could go wrong and almost certainly need a load of de-bugging software updates. The subscription model is the best way to guarantee long-term security, it works for business, which is why it works for us too.

What incentive do they have to look after you after that initial sale? What happens if the box fails or fire/theft/flood? What happens if they go bust? When they've got you by the balls, what's to stop them imposing a steep subscription? If the tech already exists, why has nobody done this before?
 
Well, as I say, this tech is available commercially, the easiest way to buy into it is AWS Durable Storage, I think Glacier is the one (Amazon's not the company I'm involved with)

The main usage is for archive storage, the various commercial offering all have in common 99.999999999% or greater durability, the number of nines is only restricted by how long the offering has been available for (ie they can't claim it is more reliable than the number of hours it has been running). This makes it particularly suitable for long term storage retention and recovery, one of the major problems for regulated companies that might have to keep data for 40+ years. As a bonus because it is intrinsically secure making it very suitable for the sectors that need to retain data (as I explained it is impossible to decode from any single site, someone could nick a whole stack of disks, hook them up to the most powerful computer and no amount of brute force would get any coherent data). Search for 99.999999999% data reliability and you'll see a range of products from all the big players.

What they have done technically to create the offering is certainly interesting, but I can't see that it has any serious mileage in its current incarnation.
 
Last edited:
The idea of offsite distributed raid type backup which automatically syncs is very attractive IMO. You don't get that with standard cloud storage. I can come close by backing up to my server which also does automatic backups, but it's still not protected against server failure (AFAIK)... the backups are stored in a local folder.

But I do doubt that there will be enough participation to make it really functional.
 
Last edited:
Ham said:
Well, as I say, this tech is available commercially, the easiest way to buy into it is AWS Durable Storage, I think Glacier is the one (Amazon's not the company I'm involved with)
Click to expand...
As best I can figure, Glacier would cost me at least $40/mo for 4TB backup storage... that's a pretty high cost comparatively.
 
Damn right, and in general terms you get what you pay for. The interesting thing about the kickstarter is that it opens the way for grid storage, whether or not there's a practical flavour, who knows, my personal view is that you would need to compensate for the much higher rate of non-availability of devices as compared to compensating for disk failure in a commercial context. (ie, total storage will be 5x the actual data or greater)

Here's a good explanation of the underlying tech if you're interested https://www.backblaze.com/blog/cloud-storage-durability/

ETA and backblaze are $5/month
 
Last edited:
You must log in or register to reply here.
Back
Top