intelligent selection of forwarders?

[Johan Ihren]

Jim Reid <jim at rfc1035.com> writes:

> >>>>> "James" == James Ralston <qralston+ml.bind-workers at andrew.cmu.edu> writes:
> 
>     James> Honestly, we don't care a whole lot about the
>     James> load-balancing aspect of this feature (i.e., BIND finding
>     James> the best forwarder to use).  What we *do* care about is
>     James> having rapid failover from a failed forwarding server.
> 
> Er, I might be asking the obvious question but why would you *ever*
> configure a name server to forward queries to an unreliable target?
> Wouldn't it be a lot simpler to just get rid of forwarding altogether
> and have your name server find out the good and bad name servers for
> itself by following NS records?

Show me any single target that is entirely reliable ;-)

There is no such thing, obviously, which is why it is a very
reasonable request from people stuck in forwarding-hell for one reason
or another to have reasonable failover behaviour.

That forwarding-hell is indeed a bad place to be is a different
issue...

Johan