How do you calculate mysql max_connections ?
What do you take into consideration ?
Database Administrators Stack Exchange is a question and answer site for database professionals who wish to improve their database skills and learn from others in the community. It only takes a minute to sign up.Sign up to join this community
Going to post this as an answer, with the relevant information. The basic formulas are:
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_size, innodb_log_buffer_size, innodb_additional_mem_pool_size, net_buffer_size, query_cache_size
Thread Buffers: sort_buffer_size, myisam_sort_buffer_size, read_buffer_size, join_buffer_size, read_rnd_buffer_size, thread_stack
WARNING! This is from 2011 using MySQL 5.1.x. Use at your own risk
---- ORIGINAL POST ----
Here's another alternative formula in stored procedure 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.
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.