[TYPO3-core] RFC #9729: Ship optimal TYPO3 configuration by default

[Clemens Riccabona]

> 
> Clemens Riccabona wrote:
> > This is a problem, yes, but keep in mind: Too much .htaccesses will
cause a
> > slowdown of apache (and most likely any other webserver software).
> 
> Did you try to measure if it actually slows down anything? ;) Disks are
fast,
> CPUs are fast, every OS has several levels of file system cache these
days. A
> think a single database query will take longer than all .htaccess files
> together :)

First of all: I did mention it, not because two or three more .htaccess
files are slowing down apache!
But if we turn off indexing for every this and that, and then, someone has
the idea with denying access for specific filetypes via htaccess, again for
every this and that, you will end up with an htaccess in every single
directory, with several commandos, which normally should be configured
within the webservers default configuration.
Not everything that CAN be done with htaccess SHOULD be done with it.
And denying indexes via htaccess has, in my opinion, nothing to do with a
'optimal configuration' of a cms-software.
This is part of the hoster. IMO.

And now your questions:
I didn't measure it exactly, but it has effects. Yes.
But this was not on a run-of-the-mill webserver which is twiddling its
thumbs whole day long,
it was on a server with about 100 independent t3-installations, and more
than 100k fe-users a day (awstats, which sometimes counts bots also as
users). (apache2 on an openSuSE 9.3 AFAIR)

Most sources for performance optimization of apache have a similar advice
in.
e.g.: http://httpd.apache.org/docs/2.0/misc/perf-tuning.html 
The information provided in this article actually tells ME, to keep my
AllowOverride directive as slim as possible
(which it is now for years), and even the DirectoryIndex directive.

And yes again, it depends on how much webs/vhosts and users you have on your
server.
I agree, that the db-queries are much slower, and most of the time when
hosting T3 (with more than a couple of users a day) you get the bottleneck
with the IO-buffer of your hdds (which is caused by the relatively
file-system-expensive db-queries (if ram is too small, or your db to big)).
And yes, you can avoid this bottleneck by buying more RAM or 15k disks, and
pull them together to an 1-0 RAID, clustering and so on.

But why should someone buy more, more expensive hardware, if it can be
avoided?
And sometimes websites are reaching dimensions, that the hardware will be
really expensive.

To came to the point:
Many small atoms make a fat cow. ;)

Greetz 'n good n8, clemens