On Thu, Mar 08, 2007 at 14:05:15 +0100, Felix von Leitner wrote:
> Thus spake Sami Farin (safari-dns@safari.iki.fi):
> > Well BIND seems to do better, according to this.
> > http://www.lurhq.com/cachepoisoning.html
>
> Someone else already proved you wrong on this.
No, I was talking only about the PRNG analyzed by the vseq program.
...
> > And epoll was not the only improvement.
>
> It is not even clear whether it is an improvement at all. On my
That's because you are too stupid to realize that if it used 0.01% CPU
earlier, you won't notice if it uses now 0.005% CPU.
Everybody else understood this simple thing.
You need extra reading, for example:
http://www.kegel.com/c10k.html
> production servers dnscache produces negligible load in the first place.
--
pgpj9BZWn6aot.pgp
Description: PGP signature
|