Can a decentralised storage service be fully decentralised? Like in a cryptocoin network, where participants can run their own nodes and pow validations?
How could Storj achieve that if it’s even possible?
I imagine something like… everyone could run a copy of the satellites address books, along with storage nodes, and Storj dev team would have only the developer role, and get a percent from all transactions… and maybe everyone could vote with their storage new changes.
I lack the deep knowledge of the inner workings of Storj, and for now, I think this is an utopic system, but maybe you have some sugestions and ideeas.
satellite is not only a book, it control that no one gambling system, when client download it ask from satellite, and when your node report about client download, satellite know that client asked this data. there is more security layers i think, some keys, that client cand ask directly frome nodes without this keys.
You should definitely read this
1 Like
The Web3 version of decentralization is focused on user control, control of the protocol, control of prices, control of the transaction; to be clear, no one entity has ultimate control, but the community (users) do.
A simple test of the degree of decentralization is the reliance on the entity, such as Storj. If Storj disappeared would the network continue uninterrupted?
1 Like
There are several problems, which need to be solved:
- consensus ( =coordination) between satellites, this is mean - slowing down every single operation with objects; no one was able to solve this issue so far;
- the distributed encrypted database in the trustless network with speed not worse than the distributed database in the trusted environment; no solution so far too;
- this solution should be scalable; perhaps only zkSync solved this issue so far;
So, you always will solve a triangle, where you can took only two (see The Electric Car Example Applied to Decentralized Cloud Storage), making the remaining worse in the same time:
- speed
- decentralization
- trustless
See also:
- A.3 Why we’re avoiding Byzantine distributed consensus
- Coordination Avoidance On The Storj Network
- Sharing Space for Fun and Profit—Part 2
And @Vadim is correct, the satellite is also auditor, repair and payments processor, not only the address book. All three are highly depends on the database with metadata, only the address book could be extracted relatively easy (also may be not, because it need an access to the metadata, to know where nodes with the pieces of each segment of the customer’s files), but inevitably will increase the TTFB.
1 Like
The question would be if that is really appealing for customers. Especially when it comes to business customers like enterprises when they store their data on a network without a entity like Storj with some degree of central power over the network, I am not sure if this is something a company would store their data on.
3 Likes
why, Enterprise like to work with other company’s, who has control of its product and responsibility for it. It is more problem with some foundations, that not responsible for anything.
2 Likes
Yeap, I guess this makes sense; companies want a partener who can sign the papers and be accounted for if the sh.t hits the fan. So, with no one in charge to deal an agreement with, they can’t trust their data to a decentralised network.
2 Likes
I think you misunderstand. I see large organizations that are actively pursuing decentralized storage. They are concerned that the hyperscalers mishandle their data as well as lose data or allow corruption of data.
My experience is that those organizations will work with those companies that front end the decentralized storage network and provide legal agreements, SLAs and other assurances as well as pricing. [This is not the Storj model however.]
I guess it depends on where you start to decentralize. Initially the data comes from a source. Be it the person typing it in, or some CPU generating data. At which point you can store it locally, or distribute it.
You have choices in how you distribute that data. Keep a copy where you are at, send a copy to a cloud provider, or use multiple cloud providers. It all depends on your needs and how much you have budgeted for this purpose.
From the Storj view, you could create your own Satellite, then pay node operators to store your data. At that point, the single point of failure is your Satellite. You have options on backing that up as well. So the data is fully distributed and you are in control of it. But you and your company remain the point of failure.
Even if you found a different model, it would likely require payment and at some point the money runs out and the data goes with it. You would need a distributed network that was free and didn’t have a single point of failure. Maybe like Gordon’s now defunct ORC network. A socialized distributed network. Not sure how successful that would be though. I tend to think you need a company like Storj handling legal issues, compliance, education, features, onboarding, and the like. Without that it wouldn’t be considered a serious tool compared to the Enterprise storage solutions that are out there. At least not without creating a Foundation type structure that has financial backing like other major open efforts.
Just my two cents. I think Storj will evolve to be fully distributed in terms of software, over time. But as mentioned there is always a failure point it just depends how far up you go in the hierarchy.
Please note that Storj does offer SLAs to our customers.
3 Likes
Yes, but that is the opposite of decentralization.
You either need an entity to deal with and fastest decentralized storage, or do not have a such entity but have a slow fully decentralized storage in the best case, which will be either unencrypted and/or not reliable or requires to run a full node and being slow.
Your data is not stored in the one central point like with legacy cloud providers. Yes, some legacy cloud providers offer a premium option to enable a replication across several datacenters in your expense, but Storj distributes pieces of your encrypted data across the globe by default.
With a legacy cloud providers your data will be stored unencrypted unless you encrypt it yourself, in Storj the encryption enabled by default. Even your metadata is encrypted, unlike legacy cloud providers.
And you also can run your own satellite, as mentioned by @Knowledge
Google shards and encrypts your data across data centers around the world. This is still not decentralization.
In a decentralized system, there are independent entities that perform a variety of services without centralized control. Two examples that come to mind are Filebase (they use Sia or IPFS based on client choice) and Linix.eu Instead of a central authority holding your encryption keys they can perform encryption for you using your public keys.
The ultimate test of a fully decentralized system is that any entity can disappear and the customer data remains safe and available. This in fact exists today without the need for a central authority.
this either server-side encryption or hard to implement and enable, while in Storj it’s working out of the box.
see
slow with full node or unencrypted and unreliable with optional replication, as I said earlier. This is the point. I do not know about Linix, it’s unknown in decentralized storage market.
If you are religiously believe to the full decentralization, you will choose either full node (manage your Metadata yourself, inability to use on mobile or for web server without central point any kind, slow access because of consensus) or the second option with unreliability and public access integrated to the system (your files may vanish in any time, even if you pay for them, no encryption, full replication, operators have full access to your data, no privacy).
For the info - if you would setup your own Satellite and attract Nodes Operators, you will get the same level of decentralization like SIA, but without running a blockchain, and thus it will be fast. However now your satellite will be a central point of failure for your access to data (exactly like in SIA).
By the way, if you did not know, that if you lose a Metadata for the SIA node - you will lose an access to your data without possibility to recover, unless you have a backup (for example - on Storj DCS, because SIA Metadata is not encrypted), then now you know that the reliability of access in the “full decentralized” network is the same, as the reliability of your full node with Metadata. But data will still be stored on nodes while you pay.
So no silver bullet so far, unfortunately.
I hope, that three mentioned problems would be solved in the future, and we, as a Community, may suggest something to solve them.
So, for the time being, the decentralisation could be achivable, but nobody wants it.
1 Like
Seems so… But I come here exactly because of decentralization and “my keys - my data” in a first place…
I have tried many other projects before come to here. For me personally Storj is still the best of two worlds in my opinion.
3 Likes