We receive many requests for help with server stalls. They come under various names: lockup, freeze, sudden slowdown. When something happens only once or twice a day, it can be difficult to catch it in action. Unfortunately, this often leads to trial-and-error approaches, which can drag on for days (or even months), and cause a [...]
Conflict Avoidance with auto_increment_increment and auto_increment_offset
January 12, 2011 By 2 Comments
A lot of people are running MySQL Master-Master replication pairs in Active-Passive mode for purpose of high availabilities using MMM or other solutions. Such solutions generally have one major problem – you have to be very carefully switching writes as if you do not do it atomically (such as some scripts continue to write to [...]
MySQL Partitioning – can save you or kill you
December 11, 2010 By 44 Comments
I wanted for a while to write about using MySQL Partitioning for Performance Optimization and I just got a relevant customer case to illustrate it. First you need to understand how partitions work internally. Partitions are on the low level are separate table. This means when you’re doing lookup by partitioned key you will look [...]
Tuning InnoDB Concurrency Tickets
May 24, 2010 By 5 Comments
InnoDB has an oft-unused parameter innodb_concurrency_tickets that seems widely misunderstood. From the docs: “The number of threads that can enter InnoDB concurrently is determined by the innodb_thread_concurrency variable. A thread is placed in a queue when it tries to enter InnoDB if the number of threads has already reached the concurrency limit. When a thread [...]
A workaround for the performance problems of TEMPTABLE views
May 19, 2010 By 7 Comments
MySQL supports two different algorithms for views: the MERGE algorithm and the TEMPTABLE algorithm. These two algorithms differ greatly. A view which uses the MERGE algorithm can merge filter conditions into the view query itself. This has significant performance advantages over TEMPTABLE views. A view which uses the TEMPTABLE algorithm will have to compute the [...]
When the subselect runs faster
March 18, 2010 By 9 Comments
A few weeks ago, we had a query optimization request from one of our customer. The query was very simple like:
1 | SELECT * FROM `table` WHERE (col1='A'||col1='B') ORDER BY id DESC LIMIT 20 OFFSET 0 |
This column in the table is looks like this:
1 | `col1` enum('A','B','C','CD','DE','F','G','HI') default NULL |
The table have 549252 rows and of course, there is an index on the col1. MySQL estimated the cardinality of that index as [...]
Introducing percona-patches for 5.1
February 9, 2010 By 21 Comments
Our patches for 5.0 have attracted significant interest. You can read about SecondLife’s experience here, as well as what Flickr had to say on their blog. The main improvements come in both performance gains and improvements to diagnostics (such as the improvements to the slow log output, and INDEX_STATISTICS). Despite having many requests to port [...]
InnoDB: look after fragmentation
November 5, 2009 By 12 Comments
One problem made me puzzled for couple hours, but it was really interesting to figure out what’s going on. So let me introduce problem at first. The table is
1 2 3 4 5 6 7 8 | CREATE TABLE `c` ( `tracker_id` int(10) unsigned NOT NULL, `username` char(20) character set latin1 collate latin1_bin NOT NULL, `time_id` date NOT NULL, `block_id` int(10) unsigned default NULL, PRIMARY KEY (`tracker_id`,`username`,`time_id`), KEY `block_id` (`block_id`) ) ENGINE=InnoDB |
Table has 11864696 rows and takes Data_length: 698,351,616 bytes on disk The problem is that after restoring table from mysqldump, the query that scans data [...]
How (not) to find unused indexes
October 16, 2009 By 2 Comments
I’ve seen a few people link to an INFORMATION_SCHEMA query to be able to find any indexes that have low cardinality, in an effort to find out what indexes should be removed. This method is flawed – here’s the first reason why:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 | CREATE TABLE `sales` ( `id` int(11) NOT NULL AUTO_INCREMENT, `customer_id` int(11) DEFAULT NULL, `status` enum('archived','active') DEFAULT NULL, PRIMARY KEY (`id`), KEY `status` (`status`) ) ENGINE=MyISAM AUTO_INCREMENT=65691 DEFAULT CHARSET=latin1; mysql> SELECT count(*), status FROM sales GROUP by status; +----------+---------+ | count(*) | status | +----------+---------+ |   65536 | archived | |     154 | active | +----------+---------+ 2 rows in set (0.17 sec) mysql> EXPLAIN SELECT * FROM sales WHERE status='active'; # query 1 +----+-------------+-------+------+---------------+--------+---------+-------+------+-------------+ | id | select_type | table | type | possible_keys | key   | key_len | ref  | rows | Extra      | +----+-------------+-------+------+---------------+--------+---------+-------+------+-------------+ | 1 | SIMPLE     | sales | ref | status       | status | 2      | const | 196 | Using where | +----+-------------+-------+------+---------------+--------+---------+-------+------+-------------+ 1 row in set (0.06 sec) mysql> EXPLAIN SELECT * FROM sales WHERE status='archived'; # query 2 +----+-------------+-------+------+---------------+------+---------+------+-------+-------------+ | id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra      | +----+-------------+-------+------+---------------+------+---------+------+-------+-------------+ | 1 | SIMPLE     | sales | ALL | status       | NULL | NULL   | NULL | 65690 | Using where | +----+-------------+-------+------+---------------+------+---------+------+-------+-------------+ 1 row in set (0.01 sec) |
The cardinality of status index is woeful, but provided that the application [...]
High-Performance Click Analysis with MySQL
December 22, 2008 By 13 Comments
We have a lot of customers who do click analysis, site analytics, search engine marketing, online advertising, user behavior analysis, and many similar types of work. The first thing these have in common is that they’re generally some kind of loggable event. The next characteristic of a lot of these systems (real or planned) is [...]
