Bind 9.21: no more recursive clients

[Simon Waters]

"Sam Hayes Merritt, III" wrote:
> 
> On Mon, 8 Jul 2002, Simon Waters wrote:
> 
> > I think the default limit is the same, but BIND 9 is
> > substantially slower on single CPU machines, so maybe that is
> > why you hit the limit?
> 
> ?
> 
> Any performance stats to correlate this?

Only the stuff Rick Jones has posted previously, mostly for
authoritative servers. I don't think Nominum, or anyone else,
are disputing it. BIND 9 does more sanity tests so that must
cost something even if it was coded similarly to 8. The last
reports I saw suggested the gap was closing, but it was still
nearly a factor of 2.