On Tue, Oct 17, 2000 at 07:21:43AM +0200, Ralf S. Engelschall wrote:
>
> Sure, shmcb is not really faster in usual storing. But it is more
> efficient when the session cache filled. Then shmht has to perform
> expensive expire operations which should cost a lot more than what shmcb
> does. So you have to measure over a longer time.
Right, what I figured.
> > But the difference between the old/new build
> > is huge and way more than I expected! (But hey, I'll take a 65% speed boost
> > for free anyday!)
>
> I've still no clue why this is the case. There should be no much
> difference for the shmht stuff, because between 2.6 and 2.7 for
> shmht mainly the abstraction layer was added. And this is a harmless
> additional function call indirection. So I really cannot image why shmht
> now should be slower than before.
Actually, the new build was much faster than before.
> If there is a difference, I guess it is more in OpenSSL. Can you
> retry shmht with Apache/1.3.12+mod_ssl/2.6.6+OpenSSL/0.9.6 vs.
> Apache/1.3.14+mod_ssl/2.7.1+OpenSSL/0.9.6?
This was going to be my second test. My only thought was that openssl
0.9.6 is a lot faster than openssl 0.9.5a. I'll see if I can squeeze in
the time to benchmark Apache/1.3.12+mod_ssl/2.6.6+OpenSSL/0.9.6 on the
same machine.
There's nothing obvious in the CHANGELOG for openssl 0.9.6, so this will
be interesting.
-Dave
______________________________________________________________________
Apache Interface to OpenSSL (mod_ssl) www.modssl.org
User Support Mailing List [EMAIL PROTECTED]
Automated List Manager [EMAIL PROTECTED]
