On Tue, 2004-05-18 at 02:16, Paul Vixie wrote: <SNIP> > If you'd like to unify something, perhaps it could be DNS client behaviour > and network-owner recursive caching forwarder design. And while you're at > it, please outlaw those fiendish DNS-based load balancers. f-root should > still be a 486DX2-66 like it was in ~1995, rather than fifty 1GHz pentiums, > and the 500X load 10 years later is due to client stupidity, not population > growth or backbone speed increases. Paul, and other rootserveroperators (good scrabble word :), what would your answer/problem/arguments/... be if an ISP would decide to inject routes to the root-servers into their local network and point these request to a local dns cache(s), which would have the correct routes to the the global rootservers of course. It is of course kind of hijacking the connectivity what is happening here. And might this be an idea for a BCP for ISP's? Or another thought that have been raised recently on the 6bone list: Would it be an idea to have 2+ independent globaly routable prefixes, thus in IPv4 2x at least /24 and in IPv6 2x /32 which are allowed to be anycasted by anyone, just like the 6to4 stuff currently. So that ISP's could point these prefixes to their local dns caches, similar to the above but: documented which prefixes those are and no evil hijacking. This could also allow for DNS-client to have hardcoded addresses of these caching DNS prefixes lightening the load on the root servers as with anycast you will always get an answer from the closest one, if all is well and murphy is on his day off of course ;) Greets, Jeroen
Attachment:
signature.asc
Description: This is a digitally signed message part