Ive been an SNO for almost a year, having a fair bit of gear invested in storj.
But depending on now existing satelites it is not truely decentralized and not 100% bulletproof, a well orchestrated ddos could bring all the satelites down for days.
When will we be able to host our own satelites that customers can use? I know its in planned but havent heard any update about it.
Cheers.
Hi! The answer probably depends on what exactly you mean by ‘truly decentralized’. Different people mean different things when they talk about decentralization.
Going by the definition used on Wikipedia here, the Storj network and protocol are already entirely decentralized. There is no single centralized authority making decisions for all peers, and all peers make local autonomous decisions toward their individual goals. Storj is decentralized in the same way as Bittorrent: yes, there are points of coordination shared by multiple peers (in Bittorrent, these are the trackers), but anyone can set up their own point of coordination and join the network.
You are presumably using a more restrictive definition of decentralization, but it’s not clear which one, so you might need to be more specific on that point. You mention an orchestrated DDoS, so perhaps you want a system with no single points of failure? But existing satellites might satisfy that as well. Satellites are not single servers. A satellite is really just a trust boundary—so in practice they are many coordinating servers accessed through redundant load balancing.
And yes, anyone can set up their own satellite. The software is open source, which would make it hard for us to stop other satellites from existing even if we wanted to! But it can be nontrivial to host any application in a robust, reliable, and available way, so it takes some effort and resources.
if ever it will be posible, this satelite made by you will be as GrolaG sattelite as you will be responsible for it and pay money to SNO for used space, take money from clients, Sattelite operator is responsible for all this. Are you redy for this? You will need big computing resorses and bandwidth for all this, Enterprise level hardware and software, 24/7 support, this is not game for home users.
It may never be an option for third-party satellites to become branded as Tardigrade; with that brand we are most concerned with providing satellites with rigorous uptime/durability/performance guarantees. It would not be a great business decision to trust a third party to provide that for us without a strong business relationship in place.
That is a requirement of the Tardigrade brand, not the Storj network. The network can exist without Storj Labs or Tardigrade. It’s not clear which one exactly you’re trying to disparage here.
Let’s pretend somebody has set up their own satellite - how can I make my node (linux, docker) connect to it?
There is a flexible system for adding or blocking specific satellites, or even subscribing to lists of satellites published by other people you trust! See https://github.com/storj/storj/blob/3bfb0a52462864337083a0883d11b08c0a6f0d01/docs/blueprints/storage-node-satellite-selection.md for some info.
You can’t without beeing recognized as a good and trusted satellite operator by tardigrade as @thepaul mentioned.
What you can do is basically building another tardigrade. Your own network with your own storjnode client preconfigured for your own satellites. Your own website where customers join and pay for their storage, and so on. Good luck with that.
A step in the right direction would be providing documentation for setting up a satellite.
I don’t care for this kind of decentralization. Another satellite run by some guy/organization I might not even trust doesn’t interest me too much at the moment. And you have to always ask yourself what the benefit is. Like for customers why would they choose a non-tardigrade satellite that might be less reliable? Is it cheaper? If so, they’ll surely pay the SNOs less too so why should I as a SNO participate in that satellite.
Not even Tardigrade satellites are trusted by ALL nodes. Nodes are autonomous and expected to act in their own rational self-interest. We can’t force them to trust anything.
If you use the storagenode software as packaged by Storj Labs, then your node will trust Tardigrade satellites by default, which might be what you mean. But if you ran your own satellite or network of satellites, you could also distribute storagenode packages that trust your satellites by default.
A good point- but it’s possible that others might run good trustworthy satellites that strike a different balance between durability/availability and cost. A satellite run on a single Linode server with a local PostgreSQL instance as its database might be good enough for the needs of many users, and the significantly lower cost of that satellite could be passed on both as lower-cost storage and higher payments for SNOs.
Or the reverse: maybe the availability and reliability guarantees made by Tardigrade aren’t enough for some users, and someone runs an even more expensive satellite in fully-replicated geo-distributed nuclear bunker datacenters.
Yes, that’s definitely a valid point. We haven’t done much in that direction yet. Hopefully we can get there soon.