�ݺ�ߣ

<Insert Picture Here>
Intro to MySQL EXPLAIN
Ligaya Turmelle
Senior Technical Support Engineer – MySQL
http://joind.in/2257
Friday, October 29, 2010

MySQL Basics
• 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.
• a key part of LAMP (Linux, Apache, MySQL, PHP / Perl / Python),
the fast-growing open source enterprise software stack.
• Site: http://www.mysql.com/
• Download: mysql.org or http://dev.mysql.com/downloads/
• Online Manual: http://dev.mysql.com/doc/refman/5.1/en/index.html

Objectives
• So what are we hoping you will learn
– How to get EXPLAIN output
– How to read the output of EXPLAIN
– What does it mean? What can we do to speed things up?
– Briefly discuss indexing techniques
– Briefly discuss query optimization techniques

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

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)

EXPLAIN Output
• Each row provides information about one table
• EXPLAIN output columns:
id key_len
select_type ref
table rows
type Filtered (new to 5.1)
possible_keys Extra
key

EXPLAIN Output

select_type
• The type of SELECT
• Can have lots of values and each means something
different
– SIMPLE – normal SELECT
– PRIMARY – outermost SELECT
– DERIVED - subquery in the FROM clause that is placed in a
temporary table (table is derived from the subquery).
– UNION – second or more SELECT statement in a UNION
– SUBQUERY – a SELECT that is not in the FROM clause
• UNION and SUBQUERY can have DEPENDANT if they
use the outer SELECT.

table
• Displays the name or alias of the table used
• 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.
• Derived tables will be noted and numbered in the
listing
mysql> EXPLAIN SELECT * FROM (SELECT * FROM COUNTRY) AS Country1 LIMIT 5;
+----+-------------+------------+------+---------------+------+---------
+----+-------------+------------+------+---------------+------+---------
| 1 | PRIMARY | <derived2> | ALL | NULL | NULL | NULL
| 2 | DERIVED | COUNTRY | ALL | NULL | NULL | NULL
+----+-------------+------------+------+---------------+------+---------
2 rows in set (0.02 sec)

type
• The access type for the SELECT query
• Various methods:
PREFERENCES SELECT TYPE
BEST system/const
⇩ eq_ref
⇩ ref
⇩ ref_or_null
⇩ index_merge
⇩ range
⇩ index
WORST ALL

Access Types

system/const
• Best types
• system
– From an in memory table
– Has only one row of data
• const
– optimizer knows that it can get at most one row from the
table.
– Primary key or unique key
mysql> EXPLAIN SELECT * FROM Country WHERE Code='USA';
+----+-------------+---------+-------+---------------+---------+---------
+----+-------------+---------+-------+---------------+---------+---------
| 1 | SIMPLE | Country | const | PRIMARY | PRIMARY | 3
+----+-------------+---------+-------+---------------+---------+---------

ref
• Similar to eq_ref but it can return more then one row
• Selected when
– Either leftmost part of a unique key is used
OR
– A non-unique or non-null key is used
mysql> ALTER TABLE City ADD INDEX (district);
Query OK, 4079 rows affected (0.09 sec)Records: 4079 Duplicates: 0
mysql> EXPLAIN SELECT * FROM City WHERE District = 'California';
+----+-------------+-------+------+---------------+----------+--------
+----+-------------+-------+------+---------------+----------+--------
| 1 | SIMPLE | City | ref | District | District | 20
+----+-------------+-------+------+---------------+----------+--------

index_merge
• This is the only access type that allows you to use 2
separate indexes for a table with MySQL
• 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'G
** 1. row **
id: 1
select_type: SIMPLE
table: City
type: index_merge
possible_keys: PRIMARY,District
key: PRIMARY,District
key_len: 4,20
ref: NULL
rows: 9
Extra: Using union(PRIMARY,District); Using where

range
• Access method for a range value in the where clause
(<, <=, >, >=, LIKE, IN or BETWEEN)
– With LIKE you can use the range access only if the first
character is not a wild card character
– 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
mysql> EXPLAIN SELECT * FROM City WHERE population>10000000;
+----+-------------+-------+-------+---------------+------------+---------
+----+-------------+-------+-------+---------------+------------+---------
| 1 | SIMPLE | City | range | Population | Population | 4
+----+-------------+-------+-------+---------------+------------+---------

index
• Doing an index scan
• Seen when
– Getting data is slow...
– The data being sought is available through the index
• Different then “Using index” in Extra
mysql> EXPLAIN SELECT id FROM CityG
** 1. row **
id: 1
select_type: SIMPLE
table: City
type: index
possible_keys: NULL
key: PRIMARY
key_len: 4
ref: NULL
rows: 4079
Extra: Using index

ALL
• Full table scan
• 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 > 10000000G
** 1. row **
id: 1
select_type: SIMPLE
table: City
type: ALL
possible_keys: NULL
key: NULL
key_len: NULL
ref: NULL
rows: 4079
Extra: Using where

Back to EXPLAIN Output

possible_keys
• Provides a list of the available indexes or NULL that
the optimizer considered for use with the query
• 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.

key
• The actual index that was used for the query or NULL
• 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.
** 1. row **
id: 1
select_type: SIMPLE
table: City
type: index
possible_keys: NULL
key: PRIMARY
key_len: 4
ref: NULL
rows: 4079
Extra: Using index

key_len
• Shows the number of bytes MySQL will use from the index
• You can use this to see if the entire index or part of an index is
being used for the query.
• 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)
** 1. row *
id: 1
select_type: SIMPLE
table: City
type: index
possible_keys: NULL
key: PRIMARY
key_len: 4
ref: NULL
mysql> DESCRIBE CityG
** 1. row *
Field: ID
Type: int(11) ← int is 4 bytes
Null: NO
Key: PRI
Default: NULL
Extra: auto_increment
** 2. row *
...

ref
• Very different from the access type “ref”
• Show which columns or constants within the index
that will be used to access the data
mysql> DESCRIBE SELECT * FROM City WHERE ID=3820G
** 1. row **
id: 1
select_type: SIMPLE
table: City
type: const
possible_keys: PRIMARY
key: PRIMARY
key_len: 4
ref: const
rows: 1
Extra:

filtered
• Since 5.1.12
• Indicates an estimated percentage of table rows that
will be filtered by the table condition.
• 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.
mysql> EXPLAIN extended SELECT * from City where district = 'Michigan';
-------+------+---------------+----------+---------+-------+------+----------+---------
table | type | possible_keys | key | key_len | ref | rows | filtered | Extra
-------+------+---------------+----------+---------+-------+------+----------+---------
City | ref | District | District | 20 | const | 8 | 100.00 | Using
-------+------+---------------+----------+---------+-------+------+----------+---------

Extra
• This is where all additional information is located
• Comments you need know about
– “Using index” - you are using a covering index (getting data
from the index rather then the table)
– “Using filesort” - manual sorting in a temporary table was
done rather then using an index for the sort
– “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)
– “Using where” - filtering outside the storage engine.

Thoughts on Indexing

Indexing
• Know your data and queries
– You always have to keep the big picture in mind
– Data changes and the right index now – may not be the right
index later
• Understand how indexes work in MySQL
– Leftmost prefix
– More indexes is not necessarily better
– MyISAM vs. InnoDB
– ANALYZE TABLE

Indexing
(con't)
• Shoot for the best EXPLAIN plan
– Use indexes for sorting
– 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
– Requires that you maintain that query over time since data
changes (can't set it and forget it)
– http://dev.mysql.com/doc/refman/5.1/en/index-hints.html
– also look into STRAIGHT_JOIN

Quick Query Tuning

Query Tuning
• Use the slow query log
– Don't guess which queries are slow - know!
– 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 for under 1 sec
– You can tell it to log all queries that do not use indexes
(log_queries_not_using_indexes)
– 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?

Query Tuning
(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
– There is a lot of information available using profiling
– http://dev.mysql.com/tech-resources/articles/using-new-query-profiler.html
• Test, Test, Test
– Keep in mind load and concurrency will affect speed

Questions?

Contact:
Email:
ligaya@mysql.com
ligaya.turmelle@oracle.com
Personal Website:
http://blog.khankennels.com
Twitter: @lig
http://joind.in/2257

�ݺ�ߣ

Explain2

More Related Content

Explain2