Hi all,
First i’m a node hoster, i love Storj idea and project, i love dashboard, i love community …
but there some inconsistencies that i don’t understand maybe someone can correct me.
Me: Hi Storj
Storj : Hi
Me : i want to be a hoster and earn some $
Storj : good
Me : i will buy some disks and start my first node
Storj : don’t …
use what you own already.
Me : Okey … i have some hard drives and used pc i will use them in a raid built to host my node and add more disks later
Storj : don’t use raid …
Storj is an redundant and decentralized storage so no need for raid or any other solution …
it’s a waste of money … run 1 node per hard drive.
Me : so i can run multiple nodes at my home ?
Storj : Yes you can
Me : But you said that Storj is a decentralized Storage
Storj : hmm…
Me : Ok let’s make some nodes and get more earnings
Storj : you will don’t get more rewards as all your nodes will be treated as one big node.
Me : What !! Explain pls …
Storj : All nodes in the same /24 subnet of public IPs treated as a one node for uploads (ingress) and as a separate ones for downloads (egress).
Me : ok as only egress is paid, so i will make more money with multiple nodes
Storj : hmm …
if you make more node on 1 external /24 ip you wonk get data faster, when first node go full, then make second node. then you have more data, and more egres, more money after some time.
Looks all good to me. I don’t see an inconsistency. You can run a single node or multiple nodes behind the same IP. The result will be the same because multiple nodes in the same location get grouped together.
The question : how it can be possible that multiple nodes will be treated as a separate ones for Egress and revenues will be same as one big node ?
Because there is no node selection for egress. Egress is dependent on how many pieces you have and how many times the piece owners want to download them.
When the owner of the object downloads the object, he gets a list of the pieces together with the nodes they are on.
I hope this clarifies your question.
It seems you maybe confuse Storj with Tor where data is passed through multiple nodes before it reaches the client. This is not the case with Storj, the client communicates directly with the storage node.
The entire magic is happening in the node selection. Let’s say you are running a big node in a small network so that you get one piece of every upload. This also means you get one download every time the customer downloads the file (technically thats wrong because a download works with a subset of the original 80 nodes). Now you add 9 additional nodes on the same IP. The node selection will group them together. You will not get more than 1 piece of the same upload. Instead one of your 10 nodes will get selected for the upload. So each node has a 1/10 chance to get selected and in total you still get the same share of the file as you would get with one big node. Again this share of the file will get you download traffic from the customer.
And yes some of you might now notice one edge case. The node selection needs to group first and than make a random selection. If it selects first and groups later it would mean multiple nodes on the same IP get additional lottery tickets. They would be limited to one win per file but buying more tickets would allow them to win more frequently than others. Don’t worry we have a test in place for that. Running multiple nodes also doesn’t increase the chance to get selected. So we hit both goals. Not more than 1 piece and also no increased chance to get selected.