NGINX Friendly URL Rewrite?

[p4guru]

Yeah the Centmin Mod install is fairly similiar in what it installs (some extra stuff not in YUM). Only Nginx, PHP-FPM + PHP extensions, Memcached, NSD and CSF are source installed while everything else including MariaDB is from YUM packages. So yes everything else can be uninstalled via YUM besides source installed stuff.

In over a decade, besides software upgrades, I can count on my fingers alone how many times I needed a clean uninstall of any software.

As to script installers, the internet is full of them - Centmin Mod is only one of many out there. Small list at http://lowendscripts.com/wiki/shell_scripts but there's plenty of others out there (predominantly Ubuntu/Debian based though) !

 

[TPerry]

Scripts... what's that? Oh yeah, it's the check list I keep on hand to make sure that I install all what I need manually.

 

[Floren]

while everything else including MariaDB is from YUM packages.

The problem is: stock repositories are not optimized for performance. I had a discussion a while ago with George and explained him the differences while suggesting him specific tuneup compile steps. He performed some extensive tests and he was blown away with the results.
http://vbtechsupport.com/715/
His quote:

Intel optimized MariaDB 5.2.5 RPM is up to 32% faster than custom MariaDB 5.2.5 ‘generic’ and up to 44% faster than official MariaDB 5.2.5 RPM

I provided him the optimized compile settings, which are present into Axivo rpms.

 

[p4guru]

Yeah read those results.. remember those were MariaDB 5.2 and 2+ years ago with gcc 4.1.2 compiler. GCC 4.8.2+ and MariaDB 5.5/10 and MySQL 5.6 are upon us and alot has changed and improved since then across all MySQL variations - Oracle, Percona, MariaDB, WebScaleSQL etc.

Have you actually benchmarked your MariaDB RPM optimized packages against the officials lately ? And have you compared how your RPM packages would fair on non-Intel processor based servers ? Would be curious of the results.

 

[Floren]

@p4guru, you can do it yourself... you have have access to all tools.
Then you will know for sure if the results are accurate. Posting myself the results is biased, no?

Just make sure you run your tests on a proper Intel processor.
Edit: I just saw your comment, why would I chose a non-Intel processor when Linux is designed to run optimally on Intel? Red Hat and Intel have been working for years together to deliver state of the art performance and reliability. Not to mention your avatar, which is an amazing processor... Test it on that.

 

[p4guru]

true Intel for life

Just thought you'd have more up to date benchmark/result numbers of your own to backup the assumption and results made 2yrs ago and that they still apply today. Assumptions aren't always correct. For example, out of box performance of MariaDB 5.2 vs 5.3 - everyone assumed MariaDB 5.3 performed better out of the box.