How do you calculate mysql max_connections ?

What do you take into consideration ?

link|improve this question
2  
Good question, and I wanted to know the answer as well. Reading through this older blog about it: mysqlhacker.com/kabir/performance/… – DTest Feb 15 '11 at 14:31
feedback

3 Answers

up vote 2 down vote accepted

Going to post this as an answer, with the relevant information. The basic formula is:

Available RAM = Global Buffers + (Thread Buffers x max_connections) max_connections = (Available RAM - Global Buffers) / Thread Buffers

To get the list of buffers and their values:

SHOW VARIABLES LIKE '%buffer%';

Here's a list of the buffers and whether they're Global or Thread:

Global Buffers: key_buffer_size, innodb_buffer_pool, innodb_log_buffer, innodb_additional_mem_pool, net_buffer_size

Thread Buffers: sort_buffer_size, myisam_sort_buffer_size, read_buffer_size, join_buffer_size, read_rnd_buffer_size

My reference is this blog post. The formula sounds right, but I have not verified the list of global vs thread buffers.

link|improve this answer
feedback

Here's another alternative formula in sproc form:

DELIMITER //
CREATE PROCEDURE sproc_show_max_memory ( OUT max_memory DECIMAL(7,4))
BEGIN
SELECT ( @@key_buffer_size + @@query_cache_size + @@tmp_table_size + @@innodb_buffer_pool_size + @@innodb_additional_mem_pool_size + @@innodb_log_buffer_size + @@max_connections * ( @@read_buffer_size + @@read_rnd_buffer_size + @@sort_buffer_size + @@join_buffer_size + @@binlog_cache_size + @@thread_stack ) ) / 1073741824 AS MAX_MEMORY_GB INTO max_memory;
END//
DELIMITER ;
CALL sproc_show_max_memory(@show_max_memory);
SELECT @show_max_memory;

I’m assuming your using a MySQL database > version 5.1.x and you’re a privileged user. But you play with the max connections by inserting whatever number you want and see the results.

link|improve this answer
I just read your formula. Pretty slick !!! BTW +1 !!! – RolandoMySQLDBA Jul 20 '11 at 20:43
feedback

I'm staring with number I'm getting from simple calulation: expected_number_of_requests_per_second * expected_average_request_processing_time * 2.

For later tuning, I'm always using monitoring system with historical data and trying to have 20% reserved in case of some peak. It's a bit more complex when you are using some connections pooling (which is usually a good idea) - then you need to monitor number of used connections in pool.

link|improve this answer
feedback

Your Answer

 
or
required, but never shown

Not the answer you're looking for? Browse other questions tagged or ask your own question.