Your IPv6 range should be treated like a mobile phone number.
-
Antranig Vartanian :freebsd:replied to Jan Wildeboer 😷:krulorange: last edited by
@jwildeboer yeah true, but the real question is: how?
-
Jan Wildeboer 😷:krulorange:replied to Antranig Vartanian :freebsd: last edited by
@antranigv A campaign to establish rules and regulations for that. Maybe also create a coop/foundation that offers personal ranges for free to every human being. Will take a lot of thoughts on making it safe and practical, but it’s definitely possible.
-
Martin Bukatovičreplied to Jan Wildeboer 😷:krulorange: last edited by
@jwildeboer I wonder why would one want do that? You can setup a dns record when self hosting a service, and for a client use case, it only makes easier for others to track you in a long term.
-
Jan Wildeboer 😷:krulorange:replied to Martin Bukatovič last edited by
@marbu You and I can setup ddns because we can. But we are the 1%. We need decentralised solutions for the 99%, IMHO. How they work in privacy respecting ways will be crucial, agreed.
-
Alyx :neocat_flag_ace:replied to Jan Wildeboer 😷:krulorange: last edited by
@jwildeboer Honestly, while I fully support the intention behind it, I still don't think it's the right approach.
IPv6 and everything that comes with it is designed with the idea that the IP address itself doesn't really matter. Clients are auto-configured, often with completely random IPs, and you're generally supposed to access an IPv6 service through a domain, not directly via the IP address.
To put it differently, if you're as dependent on an IPv6 address as you are on your mobile number, you're doing IPv6 wrong. :neocat_laugh_nervous:
Switching to another ISP should require no more effort than updating a couple of DNS records. Considering you change ISPs rarely, perhaps once a year or less, I think this is perfectly reasonable. If you use something like DynDNS, you won’t have to do any work at all.
I know this is just an idea, but I’d also like to point out that this won't be cheap. Technically, it's definitely possible with relatively little effort, but there’s no way around the need for new and additional hardware at the ISP and backbone levels. The cost for a single router would easily be four to five times the current price. Applying this internet-wide would end up costing hundreds of billions of dollars. I don't think it's reasonable to make such a massive investment just so people don’t have to update their DNS records when they change ISPs occasionally.
Again, I like the intention behind it, but I don't think this specific idea is helpful.
I suggest changing the idea to something like:
Every ISP should offer an easy one click way to get a static IPv6 prefix free of charge. :neocat_floof:
-
Jan Wildeboer 😷:krulorange:replied to Alyx :neocat_flag_ace: last edited by [email protected]
@alyx the distance from your idea of „the ISP owns prefixes that can be delegated to me“ to my idea of „I have prefixes that I can delegate to my ISP“ is smaller than you think And on the implementation level they are comparable, IMHO.
-
4censord :neocat_flag_pan:replied to Jan Wildeboer 😷:krulorange: last edited by
@jwildeboer you can already just request a personal ipv6 range from ripe afaik. just none of the residential isps i know will allow you to use it on their network
-
Jan Wildeboer 😷:krulorange:replied to 4censord :neocat_flag_pan: last edited by
@4censord Yep. We’re almost there. Needs just a little push.
-
Jan Wildeboer 😷:krulorange:replied to Jan Wildeboer 😷:krulorange: last edited by
Based on valid criticisms on privacy: IPv6 prefixes should maybe be ephemeral in a coordinated way. So your private devices can communicate but only you know your prefix at any given time. Needs more thoughts.
-
Alyx :neocat_flag_ace:replied to Jan Wildeboer 😷:krulorange: last edited by
@jwildeboer Well, not on the organizational level, you're right. I mean, you can already do that. You can request a PI (provider-independent) IPv6 prefix from RIPE (through a sponsor) and then go to an ISP and ask them to announce it for you. (You probably won't find any home ISPs doing this, but various hosting companies do, for an additional fee, of course.)
But on the global implementation and technical level, the differences are enormous. The ISP has a large prefix, which they then subdivide into smaller ones internally.
If every user or customer were to delegate a prefix to their ISP, there would be no internal subdivision. Every single customer would have their own prefix, which can't really be aggregated into a larger one.
For example... The entire internet right now consists of about 1 million separate routes.
If every person or household were to have their own prefix, that would be 40 million just in Germany alone. Worldwide, we'd end up with around 5 billion. Probably even double that, since many people have multiple internet connections (home, mobile, office, etc.).
It's a huge difference if a router has to make decisions based on 1 million routes versus 5 to 10 billion routes.
Even if we ignore all the financial and technical hurdles that come with this, we still have to morally question whether the additional power consumption needed just for the computing is justified for what we would gain from it.
I honestly still think that the domain level is a sufficient enough provider independence which doesn't struggle with this issues due to the partially centralized nature of DNS.
-
Jan Wildeboer 😷:krulorange:replied to Alyx :neocat_flag_ace: last edited by [email protected]
@alyx Thank you for sharing your insights and thinking of ways to make a more decentralised use of IPv6 possible. I know I’m opening a complicated can of worms but I feel it is needed.
-
Jörn Frankereplied to Jan Wildeboer 😷:krulorange: last edited by
@jwildeboer why just give one ipv6 - a user can have millions (e.g. in Germany I would expect that each broadband users get a lot of IPv6 addresses at home). See also: https://datatracker.ietf.org/doc/rfc7721/ There are a couple of RFCs on privacy extensions.
-
Jan Wildeboer 😷:krulorange:replied to Jörn Franke last edited by
@jornfranke I said "range", not one single address ...