More efficient MySQL Maintenance logic

[tomp_gl]

Hi,

For the last few years I have been using a fairly basic MySQL maintenance script that OPTIMIZE’s, ALTER’s by the id field and ANAYLYZE’s, in that order, every table in my database.

I have been noticing this script is causing some bottlenecks as the loop iterates over large tables and started to research whether and/or how often these queries should be performed on a table.

From what I have read:

• I should only be running OPTIMIZE on tables where the Data_free variable from SHOW TABLE STATUS LIKE ‘$table’ is greater than 0
• I shouldn’t bother running ANALYZE on InnoDB tables at all as the accuracy of the cardinality is poor.

My question is - are these findings accurate and are there other considerations I should be taking into account?

This is the basic logic of my current script:

[COLOR=#0000BB]<?php
foreach ([COLOR=#0000BB]$tables as [COLOR=#0000BB]$table)
{
[COLOR=#0000BB]mysql_query([COLOR=#DD0000]“OPTIMIZE TABLE $table”);
[COLOR=#0000BB]mysql_query([COLOR=#DD0000]“ALTER TABLE $table ORDER BY id”);
[COLOR=#0000BB]mysql_query([COLOR=#DD0000]“ANALYZE TABLE $table”);
}
[COLOR=#0000BB]?>

Thanks

[gmouse]

Best of all: ALTER TABLE $table ORDER BY id has no influence on InnoDB tables which are always sorted by the primary key.

Personally, I would not run these queries on any table unless that table is totally cleared/rewritten often.