Re: [dnsext] Sending different DNS responses based on perceived proximity

[Jim Reid <[e-mail]>]

On 29 Jan 2010, at 11:52, Stephane Bortzmeyer wrote:

> But I do not think this issue is relevant: the IETF does not care
> *how* the name server decided to reply 203.0.113.1 or
> 198.51.100.198. In the same way that RFC 2616 does not say *how* the
> Web server decides to 301 redirect and to whom. We should stick with
> protocol issues: how to carry information from one machine to another,
> not how what the servers decide to reply.

Indeed Stephane. However this discussion began with a draft that  
suggested clients send IP addressing info with their queries. If the  
draft is to go further -- I hope it doesn't -- it will very likely  
attract other cruft. For example hooks for additional data such as hop  
count to help the authoritative server do its trickery. This sort of  
cruft would mean more protocol issues.

Earlier comments about the draft suggested adding other data for  
client identification. So it's not unreasonable to mention other  
metrics for proximity now because of the protocol issues that would be  
needed to accommodate them. ie What sorts of proximity are meant and  
how would they be represented in some sort of EDNS query. What those  
metrics are (or could be) doesn't matter in that context.