[dns-operations] The (very) uneven distribution of DNS root servers on the Internet

[Patrick W. Gilmore]

> I think a much better metric, but one that would be impossibly difficult to pin down or get data for, would be looking at the average number of hops between ISPs caching servers and their closest root server.

Hops are irrelevant.  Latency, packetloss, throughput (some people call all three "good-put") is all that matters.  And in this case, throughput is not a factor.

One can make an argument for AS Hops having some correlation with performance, but it is not perfect.

Any company that has a strong dependency on DNS has done the work necessary to put authorities in the places they are required.  They have economic / business incentives to do so.  The root operators have a very different incentive.  This makes the comment about the ISPs building to the roots, or similarly, the ISPs giving the roots an incentive to build, is likely the correct viewpoint.

-- 
TTFN,
patrick


On May 15, 2012, at 12:06 , Todd S wrote:
> On Tue, May 15, 2012 at 5:30 AM, Patrik Fältström <paf at frobbit.se> wrote:
>> On 15 maj 2012, at 11:14, Stephane Bortzmeyer wrote:
>> 
>> > Asking for fairness and equity (for IP addresses or
>> > root name servers) seem reasonable to me.
> 
> The devil is in the details. Network elements should on the Internet be distributed according to network topology. .
> 
> 
> I think this is the correct approach.  But I don't think it should be up to the root server operators to figure this out - they put root servers out there in reasonable network locations around the world. ISPs, if they care sufficiently, should be working to build their network in a manner that reduces the hops to get to one of the root nodes.  
> 
> I think a much better metric, but one that would be impossibly difficult to pin down or get data for, would be looking at the average number of hops between ISPs caching servers and their closest root server.
> 
> Another roughly similar approach would be to look at the geographic location of the root namesevers and correlate that with the population density for the associated region.  One could logically assume that if a caching server is within a certain radius of a node geographically, they are likely able to route to it (country boundaries/geography may change this, but I did say roughly).  That data is likely much more available, and may show that the root nameserver/population ratio isn't so bad, or may show where enhancement is required.
> 
> 
> _______________________________________________
> dns-operations mailing list
> dns-operations at lists.dns-oarc.net
> https://lists.dns-oarc.net/mailman/listinfo/dns-operations
> dns-jobs mailing list
> https://lists.dns-oarc.net/mailman/listinfo/dns-jobs