Tuesday, Debtember 07, 2010
Some musings on decentralized DNS
Less than twenty-four hours since I released my DNS resolver library and I'm getting requests for Ruby bindings.
Heh.
While I suspect I can write Ruby bindings, it might help if I actually, you know, knew the language first. It might be something to do over this holiday season.
Also, as a joke, I was asked if I could work on that distributed DNS replacement thing, but I think I'll pass on that.
It's not as if DNS isn't already distributed—it is, but there is a central authority overseeing domain names that also help manage the servers that run the top level of DNS—the so called root servers. So it's probably more accurate to say the project is a “decentralized DNS replacement thing” and therein lies the problem—how are naming conflicts resolved?
With the current system, it's “first come, first served” when it comes to names (way back in 1998, I tried registring spc.com, but found out that Time Magazine got there first—and they still own it; I then tried conner.com, but there was a hard drive manufactorer that owned that name; now it looks like bookseller owns that domain) and if there are conflicts, ICANN will step in to mediate the issue.
But in a decentralized DNS
system, how is this handled? I want the .area51
TLD, but (hypothetically speaking here)
so does Bob Lazar. Who wins? How
is this resolved? The government?
That's the cause of this attempt to decentalize DNS. The courts? Golden rule there (“he who has the gold, makes the rules” although it could be said that's the case now).
It's more of a political (or maybe social) issue than a technical issue, and thus, I don't see it going anywhere any time soon.