On Thu, Sep 23, 2010 at 12:39 AM, Camilo Uribe <camilo.uribe@stripped>wrote:
> On Fri, Jul 9, 2010 at 12:03 AM, Johan De Meersman <vegivamp@stripped>
> > This will mostly depend on your OS, really. Assuming you're running a
> > flavour of *nix on that box, I don't think you have to worry.
> Linux on 64-bits.
Yes, but is the Linux (and your MySQL) itself also 64-bits ? :-p You *can*
use all of your ram on a 32-bit linux with the Bigmem trick, but that
introduces quite a bit of overhead, and doesn't remove the per-process
limit. A 32-bit MySQL will simply not be able to address all that memory,
even on a 64-bit OS.
> The default installation will use all the ram it needs or do I have to
> configure something else?
The default setup will use all the RAM if it needs to - but it will do so by
allocating per-connection buffers, the default shared pool sizes are rather
For MyISAM you will need to tune query cache, key cache, read buffers, sort
buffers et cetera; for InnoDB you'll make a good start by allocating a large
chunk to the buffer pool. It may also (or may not, depending) be beneficial
to leave some ram for the OS' buffer cache, which caches filesystem blocks
and thus saves on raw reads.
More exact tuning pointers are best had by observing your database's
behaviour using a trending tool like Cacti or Munin. I've said it before and
I'll say it again, my three favourite tuning resources are a) the online
docs, b) Baron & C°'s book High Performance MySQL and c) the Performance
Tuning course. With the number of parameters you can tweak on a MySQL, it's
hard to give more than basic tips without hand-on experience with the
Bier met grenadyn
Is als mosterd by den wyn
Sy die't drinkt, is eene kwezel
Hy die't drinkt, is ras een ezel