[TYPO3-project-4-3] 4.3.0 vs. 4.2.7

[Rupert Germann]

hey Mattes,

Mathias Schreiber [wmdb >] wrote:
....
> By just thinking about it I think caching things in RAM should be faster
> than putting it into the database.

yes, that was also my first theory. 
But in real-world there seem to be so much more caches that at least in my
test scenario MySQL was able to work almost completely from RAM. 
that's exactly what I wanted to achieve by giving it:

innodb_buffer_pool_size = 512M
query_cache_size        = 128M
key_buffer              = 128M
tmp_table_size          = 64MB
sort_buffer_size        = 32M
myisam_sort_buffer_size = 32M
(binlog disabled)

During the tests I saw that my machine still had ~4 GB of free RAM, means
the OS uses it as appcache, the harddisc has 16MB cache and even the CPU
has 12MB L2 cache. 
So there are much possibilities to prevent harddisc activities.

It seems to be obvious that a real RAM cache (especially one on a dedicated
Server) will drastically improve the situation when the mysql Server has
many harddisc I/Os and can't work from RAM anymore.
but you could also use this additional Server as mysql slave....

these are interesting scenarios to think about but I doubt that the majority
of the TYPO3 users will benefit from this. I'd say 90% of the existing
TYPO3 sites will never reach a league where using memcached will be a
advantage. 

> But if the real numbers are different... I get worried.
> Rupi, wanna do a profiling session sometime soon?

count me in :-)

greets
rupert