![Bend SQL to Your Will With EXPLAIN Ligaya Turmelle MySQL Support Engineer [email_address]](https://image.slidesharecdn.com/explain-090521153007-phpapp01/85/Explain-1-320.jpg)
![Start at the beginning Syntax: EXPLAIN [EXTENDED] SELECT select_options Basically you just put an EXPLAIN before a SELECT statement. What does EXPLAIN do? displays information from the optimizer about the query execution plan. EXTENDED with SHOW WARNINGS Works only with SELECT statements Some statements can be converted to SELECTS – but the statement has to “touch” all the same columns](https://image.slidesharecdn.com/explain-090521153007-phpapp01/85/Explain-12-320.jpg)
Explain
- 1. Bend SQL to Your Will With EXPLAIN Ligaya Turmelle MySQL Support Engineer [email_address]
- 2. MySQL is the world's most popular open source database software, with over 100 million copies of its software downloaded or distributed throughout its history.
- 3. a key part of LAMP (Linux, Apache, MySQL, PHP / Perl / Python), the fast-growing open source enterprise software stack.
- 7. Objectives So what are we hoping you will learn How to get EXPLAIN output
- 8. How to read the output of EXPLAIN
- 9. What does it mean? What can we do to speed things up?
- 10. Briefly discuss indexing techniques
- 11. Briefly discuss query optimization techniques
- 12. Start at the beginning Syntax: EXPLAIN [EXTENDED] SELECT select_options Basically you just put an EXPLAIN before a SELECT statement. What does EXPLAIN do? displays information from the optimizer about the query execution plan. EXTENDED with SHOW WARNINGS Works only with SELECT statements Some statements can be converted to SELECTS – but the statement has to “touch” all the same columns
- 13. Output mysql> EXPLAIN SELECT c.Name FROM City c WHERE District='Florida'\G *************************** 1. row *************************** id: 1 select_type: SIMPLE table: c type: ref possible_keys: District key: District key_len: 20 ref: const rows: 15 Extra: Using where 1 row in set (0.00 sec)
- 14. EXPLAIN output Each row provides information about one table
- 15. EXPLAIN output columns id key_len select_type ref table rows type filtered (new to 5.1) possible_keys Extra key
- 16. id The SELECT identifier.
- 17. Only if there are subqueries, derived tables or unions is this incremented Derived table is a subquery in the FROM clause that created a temporary table Number reflects the order that the SELECT was done in mysql> EXPLAIN SELECT * FROM (SELECT * FROM COUNTRY) AS Country1 LIMIT 5; +----+-------------+------------+------+---------------+------+--------- | id | select_type | table | type | possible_keys | key | key_len +----+-------------+------------+------+---------------+------+--------- | 1 | PRIMARY | <derived2> | ALL | NULL | NULL | NULL | 2 | DERIVED | COUNTRY | ALL | NULL | NULL | NULL +----+-------------+------------+------+---------------+------+---------
- 18. select_type The type of SELECT
- 19. Can have lots of values and each means something different SIMPLE – normal SELECT
- 20. PRIMARY – outermost SELECT
- 21. DERIVED - subquery in the FROM clause that is placed in a temporary table (table is derived from the subquery).
- 22. UNION – second or more SELECT statement in a UNION
- 23. SUBQUERY – a SELECT that is not in the FROM clause UNION and SUBQUERY can have DEPENDANT if they use the outer SELECT.
- 24. table Displays the name or alias of the table used
- 25. Read it down and you see the order of the tables in a JOIN which may or may not be the same as the order given in the query.
- 26. Derived tables will be noted and numbered in the listing mysql> EXPLAIN SELECT * FROM (SELECT * FROM COUNTRY) AS Country1 LIMIT 5; +----+-------------+------------+------+---------------+------+--------- | id | select_type | table | type | possible_keys | key | key_len +----+-------------+------------+------+---------------+------+--------- | 1 | PRIMARY | <derived2> | ALL | NULL | NULL | NULL | 2 | DERIVED | COUNTRY | ALL | NULL | NULL | NULL +----+-------------+------------+------+---------------+------+--------- 2 rows in set (0.02 sec)
- 27. type The access type for the SELECT query
- 28. Various methods:
- 30. | eq_ref
- 31. | ref
- 32. | ref_or_null
- 33. | index_merge
- 34. | range
- 35. | index
- 36. worst ALL
- 38. system From an in memory table
- 39. Has only one row of data const optimizer knows that it can get at most one row from the table.
- 40. Primary key or unique key mysql> EXPLAIN SELECT * FROM Country WHERE Code='USA'; +----+-------------+---------+-------+---------------+---------+---------+-------+------+-------+ | id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra | +----+-------------+---------+-------+---------------+---------+---------+-------+------+-------+ | 1 | SIMPLE | Country | const | PRIMARY | PRIMARY | 3 | const | 1 | | +----+-------------+---------+-------+---------------+---------+---------+-------+------+-------+
- 41. eq_ref An index lookup that will only return one row
- 42. Only used if both conditions are met All parts of a key are used by the JOIN
- 43. Table contains unique, non-nullable key to JOIN on mysql> EXPLAIN SELECT Country.name FROM Country JOIN City ON City.CountryCode = Country.Code; +----+-------------+---------+--------+---------------+---------+---------+------------------ | id | select_type | table | type | possible_keys | key | key_len | ref +----+-------------+---------+--------+---------------+---------+---------+------------------ | 1 | SIMPLE | City | ALL | NULL | NULL | NULL | NULL | 1 | SIMPLE | Country | eq_ref | PRIMARY | PRIMARY | 3 | world.City.Countr +----+-------------+---------+--------+---------------+---------+---------+------------------
- 44. ref Similar to eq_ref but it can return more then one row
- 45. Selected when Either leftmost part of a unique key is used
- 46. OR
- 47. A non-unique or non-null key is used mysql> ATER TABLE City ADD INDEX (district); Query OK, 4079 rows affected (0.09 sec)Records: 4079 Duplicates: 0 Warnings: 0 mysql> EXPLAIN SELECT * FROM City WHERE District = 'California'; +----+-------------+-------+------+---------------+----------+---------+-------+------+------------ | id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra +----+-------------+-------+------+---------------+----------+---------+-------+------+------------ | 1 | SIMPLE | City | ref | District | District | 20 | const | 62 | Using where +----+-------------+-------+------+---------------+----------+---------+-------+------+------------ 1 row in set (0.01 sec)
- 48. ref_or_null Again similar to eq_ref and ref but Allows null values
- 49. OR
- 50. Allows null conditions mysql> ALTER TABLE Country ADD INDEX (IndepYear); Query OK, 239 rows affected (0.36 sec) Records: 239 Duplicates: 0 Warnings: 0 mysql> EXPLAIN SELECT IndepYear FROM Country WHERE IndepYear = 1905 or IndepYear IS NULL; +----+-------------+---------+-------------+---------------+-----------+---------+-------+------ | id | select_type | table | type | possible_keys | key | key_len | ref | rows | +----+-------------+---------+-------------+---------------+-----------+---------+-------+------ | 1 | SIMPLE | Country | ref_or_null | IndepYear | IndepYear | 3 | const | 31 | +----+-------------+---------+-------------+---------------+-----------+---------+-------+------ 1 row in set (0.00 sec)
- 51. index_merge This is the only access type that allows you to use 2 separate indexes for a table with MySQL
- 52. There are lots of rules of when it will be used and in what conditions – intersection, union http://dev.mysql.com/doc/refman/5.0/en/index-merge-optimization.html mysql> EXPLAIN SELECT * from City where id = 5 or district = 'Michigan'; +----+-------------+-------+-------------+------------------+------------------+---------+------+------+--------------------------------------------+ | id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra | +----+-------------+-------+-------------+------------------+------------------+---------+------+------+--------------------------------------------+ | 1 | SIMPLE | City | index_merge | PRIMARY,District | PRIMARY,District | 4,20 | NULL | 9 | Using union(PRIMARY,District); Using where | +----+-------------+-------+-------------+------------------+------------------+---------+------+------+--------------------------------------------+
- 53. range Access method for a range value in the where clause
- 54. (<, <=, >, >=, LIKE, IN or BETWEEN) With LIKE you can use the range access only if the first character is not a wild card character
- 55. Not possible with hash indexes (MEMORY or NDB tables) mysql> ALTER TABLE City ADD INDEX (population); Query OK, 4079 rows affected (0.10 sec) Records: 4079 Duplicates: 0 Warnings: 0 mysql> EXPLAIN SELECT * FROM City WHERE population>10000000; +----+-------------+-------+-------+---------------+------------+---------+------+------+-------------+ | id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra | +----+-------------+-------+-------+---------------+------------+---------+------+------+-------------+ | 1 | SIMPLE | City | range | Population | Population | 4 | NULL | 2 | Using where | +----+-------------+-------+-------+---------------+------------+---------+------+------+-------------+
- 56. index Doing an index scan
- 57. Seen when Getting data is slow...
- 58. The data being sought is available through the index Different then “Using index” in Extra mysql> EXPLAIN SELECT id FROM City; +----+-------------+-------+-------+---------------+---------+---------+------+------+-------------+ | id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra | +----+-------------+-------+-------+---------------+---------+---------+------+------+-------------+ | 1 | SIMPLE | City | index | NULL | PRIMARY | 4 | NULL | 4079 | Using index | +----+-------------+-------+-------+---------------+---------+---------+------+------+-------------+
- 59. ALL Full table scan
- 60. SLOW mysql> ALTER TABLE City DROP INDEX Population; Query OK, 4079 rows affected (0.06 sec) Records: 4079 Duplicates: 0 Warnings: 0 mysql> EXPLAIN SELECT * FROM City WHERE Population > 10000000; +----+-------------+-------+------+---------------+------+---------+------+------+-------------+ | id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra | +----+-------------+-------+------+---------------+------+---------+------+------+-------------+ | 1 | SIMPLE | City | ALL | NULL | NULL | NULL | NULL | 4079 | Using where | +----+-------------+-------+------+---------------+------+---------+------+------+-------------+
- 61. possible_keys Provides a list of the available indexes or NULL that the optimizer considered for the query
- 62. Gotcha If you have tables/column with different character sets, the indexes on the tables/column may not not be available for lookup since it has to convert from one character set to another.
- 63. key The actual index that was used for the query or NULL
- 64. key and possible_keys are where you start looking at your indexing strategy We will briefly discuss optimizing your indexes a bit later in the talk. mysql> EXPLAIN SELECT id FROM City; +----+-------------+-------+-------+---------------+---------+---------+------+------+-------------+ | id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra | +----+-------------+-------+-------+---------------+---------+---------+------+------+-------------+ | 1 | SIMPLE | City | index | NULL | PRIMARY | 4 | NULL | 4079 | Using index | +----+-------------+-------+-------+---------------+---------+---------+------+------+-------------+
- 65. key_len Shows the number of bytes MySQL will use from the index
- 66. You can use this to see if the entire index or part of an index is being used for the query.
- 67. Keep in mind that some character sets can use more then one byte per character.(Ex: UTF8 can use up to 3 bytes per character) mysql> describe City; +-------------+----------+------+-----+---------+----------------+ | Field | Type | Null | Key | Default | Extra | +-------------+----------+------+-----+---------+----------------+ | ID | int(11) | NO | PRI | NULL | auto_increment | ← INT is 4 bytes
- 68. ref Very different from the access type “ref”
- 69. Show which columns or constants within the index that will be used to access the data
- 70. rows Number of rows MySQL expects to find based on statistics Can be very different then what really shows up You can update these statistics with ANALYZE TABLE but remember that each storage engine updates it's statistics with varying levels of accuracy. Ex: InnoDB statistics are based on 10 random dives rather then on all the actual data
- 71. filtered New – since 5.1.12
- 72. indicates an estimated percentage of table rows that will be filtered by the table condition.
- 73. rows shows the estimated number of rows examined rows × filtered / 100 shows the number of rows that will be joined with previous tables. displayed if you use EXPLAIN EXTENDED.
- 74. Extra This is where all additional information is located
- 75. Comments you need know about “Using index” - you are using a covering index (getting data from the index rather then the table)
- 76. “Using filesort” - manual sorting was done rather then using an index for the sort
- 77. “Using temporary” - at some point in the execution a temporary table was made. If necessary you want to keep it in RAM (tmp_table_size and max_heap_table_size)
- 78. “Using where” - filtering outside the storage engine.
- 79. Indexing Know your data and queries You always have to keep the big picture in mind
- 80. Data changes and the right index now – may not be the right index later Understand how indexes work in MySQL Leftmost prefix
- 81. More indexes is not necessarily better
- 83. ANALYZE TABLE
- 84. Indexing (con't) Shoot for the best EXPLAIN plan Use indexes for sorting
- 85. Covering indexes Index hinting is possible though highly discouraged If you know absolutely that you should be using a specific index _and_ ANALYZE TABLE is not helping
- 86. Requires that you maintain that query over time since data changes (can't set it and forget it)
- 88. also look into STRAIGHT_JOIN
- 89. Query Optimizing Use the slow query log You can set a maximum time for a query to run and if it goes over, log it (long_query_time) In 5.1 micro time is available You can tell it to log all queries that do not use indexes (log_queries_not_using_indexes)
- 90. http://dev.mysql.com/doc/refman/5.0/en/slow-query-log.html Don't be afraid of sub-selects, but use them wisely Can you write it as a JOIN and if so – does it run faster then the subselect?
- 91. Query Optimizing (con't) Be willing to consider alternate methods of doing things Instead of one big query with lots of derived tables, use a stored procedure that builds intermittent MEMORY tables that you index and work with Use Query Profiling to get the nitty gritty Available from 5.0
- 92. There is a lot of information available using profiling
- 94. Questions?
- 95. Performance Tuning the MySQL Server Ligaya Turmelle MySQL Support Engineer [email_address]