We have a site with a similar architecture:
- 6 load balanced front-ends
- 1 dedicated database server (Dual P3 1.4GHz, 2GbRAM, RH7.2)
we serve around 3 million pages/day (all pages are dynamiquely generated,
each page needs an average of 15 SQL queries).

What we have done:
- audit ALL SQL queries and make sure that they are optimised (all selects
  use indexes,...)
- optimise mysqld configuration (tune key_buffer_size, ..)
- setup replication : each front-end is a mysql slave server and replicate
  most used tables.
- modify the site so that heaviest SELECT queries are run by each front-end
  on local slave.

Result:
- on main database server: 300queries/s average (~700q/s peak)
                           load : 0.2 average, ~0.7 peak
- on each front-end : 30q/s average (60q/s peak)

We could run more requests on the slaves but since the master server load
is so low, we have postponed those optimisations.

I think you really should audit your queries first. From my experience and what
other users have reported on this list, you should expect to be able to
run several hundred queries/s with the kind of hardware you are
using.

Hope this helps
-- 
Joseph Bueno


Chavvon Smith wrote:
> We are hosting a high volume site that gets about 1 million page views a day
> on RedHat 7.3.  We currently have 3 load balanced servers on the front end
> accessing a MySQL server on the back end.  The MySQL servers is dual P3 1ghz
> with 1 GB of RAM and when the MySQL queries hit about 50 per second, the DB
> crashes and the servers is useless unless you reset the DB.  Memory is only
> at about 50% usage, but the CPU skyrockets to 100%.
> 
> The only solution we can think of is to throw a huge server at the backend
> (i.e. 4-8 processor Compaq 8500) and keep RH 7.3 or switch to Windows 2000
> Advanced Server and cluster a few dual P3 servers together.
> 
> Any other solutions to make MySQL handle a high volume site?
> 
> CS
> 


---------------------------------------------------------------------
Before posting, please check:
   http://www.mysql.com/manual.php   (the manual)
   http://lists.mysql.com/           (the list archive)

To request this thread, e-mail <[EMAIL PROTECTED]>
To unsubscribe, e-mail <[EMAIL PROTECTED]>
Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php

Reply via email to