4 ms·

Can we kill the DNS system already and move to something more decentralized?

forgottenpass 11y ago
It is decentralized, you can spin up an alternate root right now. The problem has been, and always will be namespace coordination. You can argue that ICANN could have done better. And part of me still cares, but of course ICANN fucked up the existing root. In the current business and legal climate, it's basically a foregone conclusion.
- yellowapple 11y ago
  That's not decentralization; that's just moving the single-point-of-failure. A better system would be something based on a cryptographic blockchain/ledger, like Namecoin. Thus, no reliance on a central authority to decide which domains are or aren't valid; you instead just look at the ledger (which is maintained decentrally by network participants).
  - jsmthrowaway 11y ago
    We can't even deploy IPv6. What's your plan for getting your system deployed in every Internet connected device, including ancient routers without cryptography chops and refrigerators, then transferring the multiple billions spent on the current namespace into your system without conflict?
    - lucb1e 11y ago
      > We can't even deploy IPv6 Actually, all my traffic is v6 by default using Telenet (Belgium) and XS4ALL (the Netherlands). And yes, those are mainstream ISPs. Lots of services, most noticeably Google and thus Youtube, use it by default as well. We're nowhere near 100% yet, but as the need increases we are getting there.
      - jsmthrowaway 11y ago
        As anyone who knows what you just condescendingly explained knows, anything short of 100% doesn't alleviate the problem that motivated IPv6 in the first place. Until absolutely everything is accessible via v6, we must continue to allocate v4 (or, worse, NAT it all), so we aren't really anywhere. It was a side point. I'm on dual stack Comcast, so I already knew what you're telling me, and I made the point nonetheless.
        lucb1e 11y ago
        First of all, sorry, I did not mean to be condescending. People keep repeating we can't do X because we can't even deploy IPv6 and X is an equally large or bigger change. Since I've got v6 for years and am getting it in more and more places, we're getting there alright. So I do not understand how you can say we cannot deploy v6. You are running dual stack as well, clearly we are on the way there? And yes of course we first need widespread deployment before we can turn off v4 entirely, but the fact that we both have it means that people are getting it and that we can turn off v4 at some point in the future. And as an aside, we don't actually need 100% exactly: at 99% (or something) it's not going to be cost effective to get v4 addresses for everything anymore and more stuff will become v6-only.
  - forgottenpass 11y ago
    Oh, you're talking about decentralization to mean remove any authoritative control over a zone (rather than ability to choose the delegation system), which also requires upending the resolution design. I'm largely just referring to the namespace management and rules attached. In that respect namecoin isn't a particularly interesting entrant. It doesn't really fix problems with namespace coordination unless you decide they're irrelevant next to maintaining namecoin's ideologically purity. If namecoin were to take off it would just collapse under it's own weight once squatters move in. So, outside of ICANN pushing huge and onerous changes down to their TLDs, I don't really see the utility for namecoin beyond providing resolution for services that would get their domains revoked by a registry for policy reasons or following legal orders.
natrius 11y ago
I'd rather look up an identity to find what they've published rather than looking up a name to talk to a server that will hopefully give me what they've published. Luckily, cryptography lets us do that. http://ipfs.io http://ipfs.io

It is decentralized, you can spin up an alternate root right now. The problem has been, and always will be namespace coordination. You can argue that ICANN could have done better. And part of me still cares, but of course ICANN fucked up the existing root. In the current business and legal climate, it's basically a foregone conclusion.

yellowapple 11y ago
That's not decentralization; that's just moving the single-point-of-failure. A better system would be something based on a cryptographic blockchain/ledger, like Namecoin. Thus, no reliance on a central authority to decide which domains are or aren't valid; you instead just look at the ledger (which is maintained decentrally by network participants).
- jsmthrowaway 11y ago
  We can't even deploy IPv6. What's your plan for getting your system deployed in every Internet connected device, including ancient routers without cryptography chops and refrigerators, then transferring the multiple billions spent on the current namespace into your system without conflict?
  - lucb1e 11y ago
    > We can't even deploy IPv6 Actually, all my traffic is v6 by default using Telenet (Belgium) and XS4ALL (the Netherlands). And yes, those are mainstream ISPs. Lots of services, most noticeably Google and thus Youtube, use it by default as well. We're nowhere near 100% yet, but as the need increases we are getting there.
    - jsmthrowaway 11y ago
      As anyone who knows what you just condescendingly explained knows, anything short of 100% doesn't alleviate the problem that motivated IPv6 in the first place. Until absolutely everything is accessible via v6, we must continue to allocate v4 (or, worse, NAT it all), so we aren't really anywhere. It was a side point. I'm on dual stack Comcast, so I already knew what you're telling me, and I made the point nonetheless.
      - lucb1e 11y ago
        First of all, sorry, I did not mean to be condescending. People keep repeating we can't do X because we can't even deploy IPv6 and X is an equally large or bigger change. Since I've got v6 for years and am getting it in more and more places, we're getting there alright. So I do not understand how you can say we cannot deploy v6. You are running dual stack as well, clearly we are on the way there? And yes of course we first need widespread deployment before we can turn off v4 entirely, but the fact that we both have it means that people are getting it and that we can turn off v4 at some point in the future. And as an aside, we don't actually need 100% exactly: at 99% (or something) it's not going to be cost effective to get v4 addresses for everything anymore and more stuff will become v6-only.
- forgottenpass 11y ago
  Oh, you're talking about decentralization to mean remove any authoritative control over a zone (rather than ability to choose the delegation system), which also requires upending the resolution design. I'm largely just referring to the namespace management and rules attached. In that respect namecoin isn't a particularly interesting entrant. It doesn't really fix problems with namespace coordination unless you decide they're irrelevant next to maintaining namecoin's ideologically purity. If namecoin were to take off it would just collapse under it's own weight once squatters move in. So, outside of ICANN pushing huge and onerous changes down to their TLDs, I don't really see the utility for namecoin beyond providing resolution for services that would get their domains revoked by a registry for policy reasons or following legal orders.

natrius 11y ago

I'd rather look up an identity to find what they've published rather than looking up a name to talk to a server that will hopefully give me what they've published. Luckily, cryptography lets us do that. http://ipfs.io http://ipfs.io