A while back there were inquired to if it was correct that SNO’s was only allowed to 1 IP address as according to the Terms of Service, and no satisfactory answer was given.
I think there are so fundamental issues with allowing multiple IP’s even if I’m not aware of how to actually restrict it in practice… the problem becomes that if 5% of SNO’s have 10 Internet connections on different subnets/24, then they would represent 50% of the whole… thus of the total, it would represent 1/3.
So 5% of SNO’s could have 1/3 of the network data if supplying enough disks and having 10 internet connections each… all connected to single servers… but why stop there…
You get the idea… I just would like clarification if I should get a second IP or not.
Economically it seems like the only logical next step for me.
Surprisingly the terms currently stipulate that you must use a different IP address per node!
We will be changing this.
We want there to be one wallet address per SNO for tax and legal reasons, but we have no problem with SNOs running as many Storage Nodes on as many IPs as they want. It is our job (not the SNOs) to keep the data and network balanced, and our Node selection process continues to have tweaks and improvements from time to time as we discover issues. We have not seen the last of the issues of course, and we will continue to prioritize fixes and adjustments as needed.
It’s worth noting that while we intend to change the SNO T&Cs to allow SNOs to operate more than one Node behind the same IP, the most important things to keep in mind when running multiple nodes are that:
The network balances load across Nodes and uses IP filtering as part of that process (5 Nodes behind 1 IP will get less traffic than 5 Nodes behind 5 IPs)
Each Node should have it’s own physical drive. Running multiple Nodes on virtual machines sharing the same physical storage isn’t allowed under the terms. The failure of that hardware would mean the failure of all of the Nodes using that hardware.
We strongly recommend just using one drive per Node and not investing in RAID or other hardware for redundancy. The network is built to manage redundancy on the software side.
Yep, and then as more SNOs do that, we will continue to adjust and balance the prioritization of various network routes to maximize broad distribution of data as much as possible.
Obviously the larger the network is, the easier this is for us!
We have 1240 nodes in Germany alone. If something were happen to the internet connectivity of Germany, data where all of the pieces are in Germany would no longer be available outside of Germany. This is a broad problem for us, and a big reason why we try and get as broad of distribution of data as possible. We try our best to put data on as many different internet routes as possible to avoid disaster in this way.
I would love to see a little more detail around this. It seems currently node selection for upload is still random. Are there any specific plans or changes in the works you can tell us a little more about?
Yeah as I’m sure you know, what is currently released is that we select over 100 IPv4 /24-netmasks at random for each upload, and then use the fastest 80. So far, our analysis shows that this is giving a good enough distribution, though we still have more work to do. Ultimately we have begun to discuss internally making better decisions based on (1) what is actually in nearby routing tables, (2) giving node operators the ability to report higher quality location data, or (3) trying to select a larger edit-distance set of netmasks instead of just /24 netmasks, but we haven’t implemented any of that yet.
Thank you for responses Time flies when you are having fun! This concludes the time allotted for the hosted portion of the Q&A, and we’ll be closing out the threads shortly.
I feel like that has come up before either in a past Town Hall, or else in some conversations Ive had with coworkers. @jtolio I know the QA is over, so what makes sense to me here is to split the above comment into a regular forum topic discussion.
The reason Im closingout threads here is because I dont want anyone to feel ghosted as the hosts move on to the rest of their day. But youre totally welcome to keep discussing in the main part of the forum, at whatever pace suits everyone best. Thanks
Thank you for responses Time flies when you are having fun! This concludes the time allotted for the hosted portion of the Q&A, and we’ll be closing out the threads shortly.
Time flies when you are having fun! This concludes the time allotted for the hosted portion of the Q&A, and we’ll be closing out the threads shortly.