Lancashire, Pete wrote:
I'm doing some tests on a small Sun server, a 280R to demo
MySQL vs. Oracle. For the demo I using a subset of a test table.

The destination server will be a Sun 880 with 8 1.2 GHz CPU's,
16 GB RAM, Sun T3+ RAID Array.

Any suggestions on settings or other changes would be helpful.

Also what would be the best variables/settings to compile a binary
for a server of this type ? Gcc 3.3.2, Solaris 9.

number of rows is only 10,735,291

MYD is 6.7 GBytes, and MYI is 1.4 GBytes.

This table is a subset of the real table of about 120 Gb.

the test/lab server is a small sun box with 2 each 900 MHz CPUs, and 8
GB of RAM.

I'm trying to avoid using disk until I get a better performance disk
subsystem.

The version is the website binary, solaris 64 bit.

the setting i currently have

[mysqld]
set-variable = myisam_sort_buffer_size=4096M
set-variable = sort_buffer_size=4096M
set-variable = read_buffer_size=512M
set-variable = key_buffer_size=4096M
set-variable = table_cache=1024

tmp_table_size      = 1024M
max_heap_table_size = 1024M

a search on text fields such as

select substring_index(request_uri,'/',-1) jsp, count(*) from apache_jan
where request_is_jsp = 1 group by jsp;

takes 8 1/2 minutes.

Pete:


I am afraid your question is somewhat like : "It takes really long to drive from Los Angeles to San Francisco going through Denver in my super speedy car. How can I make the car faster to cut the trip time?" Perhaps a better question to ask first is "Do I really have any business in Denver that I have to attend on my way to San Francisco?", and hope that the answer will be "no".

To apply the above to your case, if possible, rebuild the table to store the file extension ( I assume you can identify the type of request by extension), and the file stem (file name without extension) for each request. Then you can drop request_is_jsp and request_is_servet fields. Also, think about your other frequent queries to see if they can benefit from a different data layout, and consider using summary tables to avoid lengthy repeated group by queries.



--
Sasha Pachev
Create online surveys at http://www.surveyz.com/

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



Reply via email to