On Mon, Apr 22, 2013 at 04:50:52PM +0300, Kostas Zorbadelos wrote: > Stuart Henderson <s...@spacehopper.org> writes: > > > On 2013-04-19, Kostas Zorbadelos <kzo...@otenet.gr> wrote: > >> root@dmeg-dns1 ~ # /usr/local/sbin/named -V BIND 9.9.2-P2 built with > >> --enable-shared' '--enable-threads' > > > > You could try rebuilding the port without --enable-threads and see if it's > > any different. > > > > I rebuilt the port without threads > > root@dmeg-dns1 ~ # /usr/local/sbin/named -V > BIND 9.9.2-P2 built with '--enable-shared' '--with-libtool' > '--prefix=/usr/local' '--sysconfdir=/etc' '--mandir=/usr/local/man' > '--infodir=/usr/local/info' '--localstatedir=/var' > '--disable-silent-rules' 'CC=cc' 'CFLAGS=-O2 -pipe' 'CXX=c++' > 'CXXFLAGS=-O2 -pipe' > using OpenSSL version: OpenSSL 1.0.1c 10 May 2012 > using libxml2 version: 2.8.0 > > There was no great difference in performance. I could see in the > multi-threaded BIND version CPU utilization of up to 300% (Linux showed > up to 800% on the same tests). > In repeating tests with > resperf (to have a full cache) and giving more slack in the rampup time > of the queries (resperf -r 240 -s <server IP> -d <recorded_queries>) I > was able to reach around 20K queries / sec on my hardware, half the > performance of Linux on the same hardware, BIND version and > configuration. PF disabled. > > Might also test unbound. If I do, I will post the results. > > PS: I think I saw a slight improvement using -U 4 (provided by the > port's init script) >
Don't forget to increase the UDP recvbuffer space. The default is somewhat small and will result in drops. At least you should invest some time to play with that value and see if it helps. -- :wq Claudio
