This is a bit ot but relying on a public IP range for node distance is not the best idea, rather the opposite
Please, suggest your solution. We need to have each piece of the segment stored across the globe, not in a few datacenters or in the same ISP.
Because if anything would happened with the ISP or datacenter, customers data will be in dangerous.
2 Likes
I’ve worked in the ISP business for quite some time and it’s very uncommon that an ISPs whole network would die, not even when L3 had huge issues in the us that happen. IPv3 addresses are limited and the subsets has moved to class-less allocation. If you want only one node per ISP you can look for AS#
Node-identity + geoip seems better but I do not have a perfect solution
it would be just not posible to get desired space for storj. Becase people who have spare HDD they usualy have lot of them. But it very hard to put them do lot of diferent locations. And if you put behind /24 then you have low usage and dont get even power cost back.
But it is today. I hope in production it wil change, because today there is smal portion of testers that loding data. In production there will be lot of people with files, then there will lot of files and even if you get 1 peace from 1 client and 100-1000 client trafic wil be huge.
You are saying that in production, there is a greater chance that the PUT/GET will increase to a single “subnet” and therefore the revenue will increase, bearing in mind you will share the “spread” along all others on the same /24.
I guess what I wanted to say is that there is no definition of a “24-submask”, that was removed 20 years ago. A single “24” can be shared between two different ISPs, especially now when all IPv4 addresses are used. If not between different ISPs definitely between different cities or regions in the same ISP. So I could be on a /24 and another customer of the same ISP could be on the same subnet, but 1000 miles away.
Yeah, but what are the odds of this happening with only a few thousand SNO’s? Somehow there needs to be a line drawn. Or do you have a better solution of spreading all files around the globe?
And if ISP is down, both nodes in 1000 miles away are went down too.
The node should be on a different ISP. In this case it’s unlikely that they have a shared subnet.
They are both IPv4. Should your ISP support it, what if one of you uses only a IPv6 IP address?
The node will work only with customers with IPv6
Hi Alexey, what exactly do you mean with: The node will work only with customers with IPv6?
Do I understand correctly, that node in storj network has to support IPv6? I am asking, because I didnt configure IPv6 on my node and despite this my node is UP and running.
Pretty sure he meant IPv4 not IPv6
1 Like
It depends on the node configuration. If the node only has IPv6 connectivity, it will only receive data from customers who also have IPv6 connectivity. If the node has both, it can receive from both IPv4 and IPv6 customers. If the node has only IPv4, then customers with IPv4 will upload but IPv6-only customers wouldn’t be able to.
Does this make more sense?
2 Likes
I dont believe this is the case I think it only supports IPv4 not IPv6
Hi, yes, absolutely.
Only one satellite currently supports IPv6 and it has been confirmed elsewhere on the forum that IPv6 traffic does occur although not nearly as much as IPv4 traffic.
So getting back to the original reason ipv6 was mentioned, can one set up a ipv4 and a ipv6 storage node on the same home network and have them treated as completely seperate storage nodes?
It’s support both IPv4 and IPv6, but the IPv6 is working for communication only with one satellite. We doesn’t have a huge amount of customers with IPv6 only.
If your node have only IPv6, it’s able to transfer data only to/from customers with IPv6.
So better to have a dual stack. In this case your node will be able to serve customers from both networks
Yes, if you able to completely separate them and configure the right routing.
Sounds like a plan in the making.
You don’t need separate nodes to take advantage of the additional ipv6 traffic. If you have a dual stack node you get all that same traffic.