Regarding BIND9.3.2

[Stefan Puiu]

It's actually a preprocessor define - you can find the following lines
in lib/isc/mem.c:

/*
 * Define ISC_MEM_USE_INTERNAL_MALLOC=1 to use the internal malloc()
 * implementation in preference to the system one.  The internal malloc()
 * is very space-efficient, and quite fast on uniprocessor systems.  It
 * performs poorly on multiprocessor machines.
 */
#ifndef ISC_MEM_USE_INTERNAL_MALLOC
#define ISC_MEM_USE_INTERNAL_MALLOC 0
#endif

Change the #define line so the variable is defined to 1 instead, and
then recompile.

On 2/20/06, Alex Sharaz <A.Sharaz at hull.ac.uk> wrote:
> o.k. so where's the config option to use the internal malloc?
> Tried ./configure --help >config.help but nothing seems to stick out
>
>
> -----Original Message-----
> From: bind-users-bounce at isc.org [mailto:bind-users-bounce at isc.org] On
> Behalf Of Paul Vixie
> Sent: 19 February 2006 23:15
> To: comp-protocols-dns-bind at isc.org
> Subject: Re: Regarding BIND9.3.2
>
> christ lee <christfer1.lee at gmail.com> writes:
>
> > I recently came to know that new version of BIND (bind9.3.2) is
> > released.  if you have any idea about the changes made from bind9.3.1
> to
> > bind9.3.2 with respect to improve the query throughput, just give me
> > reply.
>
> As barmar said, there's a CHANGES file you can look at to find out
> what's
> new in any given version of BIND9.  Query throughput has improved
> slightly
> in 9.3.2, and we now know that using the internal "malloc" will improve
> it
> even more (that's a build parameter).  However, the real performance
> boost
> will come in 9.4.0, especially on multiprocessors.  If your only known
> reason to upgrade is performance, then you need to be aware of more good
> reasons -- see the CHANGES file -- but if you can't be convinced to just
> run-the-latest-greatest then you should plan on 9.4.0 as an upgrade
> point.
> --
> Paul Vixie
>
>
>
>