We finally split our huge logging aspect of an application to its own system, 
and we changed the table type from InnoDB to MyIsam.  This is a logging table, 
1 log writer, many, many selects from customer care reps looking up the last 
500 or so records.  These are logs of timed events, so they always need to see 
the most recent 500 records.

The memory buffers that are set for now are as follows:

### OPERATIONAL SETTINGS
key_buffer_size=416M
myisam_sort_buffer_size=128M
join_buffer_size=128M
read_buffer_size=16M
sort_buffer=256M
read_rnd_buffer_size=32M
query_cache_size=32M

The box has 4 GB, it is Linux, so I am thinking I can use up to 3 GB.  I did 
not size these initial settings but the more I see on what others use, I learn 
of all kinds of buffers I never thought of using before.  I understand that 
some of these buffers will degrade performance if too large, while others can 
grow to the sky, and only improve performance.

I would appreciate some additional input on these settings, and advice how I 
could optimize these even further to obtain as much through-put as possible 
from the system.  Are there any other buffers I could set?  The majority of all 
action are inserts from the log writer, and selects from customer care reps.

One other thing, what can be optimized if the log writer is split into multiple 
threads, and many insert processes are happening at once?

Regards,

Thomas A. Lekai
Vonage Holdings
[EMAIL PROTECTED] <mailto:[EMAIL PROTECTED]>


--
MySQL General Mailing List
For list archives: http://lists.mysql.com/mysql
To unsubscribe:    http://lists.mysql.com/[EMAIL PROTECTED]

Reply via email to