[dns-operations] Client retry behavour?

[Paul Vixie]

> > i've been thinking for about a decade that we need a blast protocol for
> > the stub resolver.
> 
> I think the assumptions that went into the creation of a 'stub  resolver'
> should probably be revisited in the face of much more  powerful machines
> and more memory.  With the exception of (some)  embedded devices and given
> the Xeno's Paradox dream of deploying  DNSSEC, I suspect it would make more
> sense for most everybody to run  a local caching server (or "full resolver"
> or "recursive nameserver").

i completely agree.  even a wristpad PDA has more memory and cpu than the vax
750's for which the current stub resolver model was designed.  and if i were
writing a recommendation, i would recommend a full resolver everywhere.  but
i am not writing that recommendation, mostly because i don't think people are
likely to change their behaviour.  so i'm thinking about how to improve the
model that people are going to keep using.  xeno's paradox would be paradise
compared to the problems we're actually addressing.