SlideShare a Scribd company logo

Optimizing Queries with Explain

MYXPLAIN, profile picture
MYXPLAIN

The document discusses how to use EXPLAIN to optimize SQL queries. EXPLAIN shows how tables are joined, indexes used, and records examined. It returns information like the number of tables, join types, and data access methods. The fastest strategies are const, which uses a primary or unique key to lookup at most one value, and eq_ref, which joins on a unique index. EXPLAIN helps identify inefficient queries to improve performance.

Technology
1 of 90
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
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
Optimizing Queries with EXPLAIN http://bit.ly/LTsbuJ Sheeri Cabral Senior DB Admin/Architect, Mozilla @sheeri www.sheeri.com
EXPLAIN  SQL extension  SELECT only  Can modify other statements: UPDATE tbl SET fld1=“foo” WHERE fld2=“bar”; can be changed to: EXPLAIN SELECT fld1,fld2 FROM tbl;
What EXPLAIN Shows  How many tables  How tables are joined  How data is looked up  If there are subqueries, unions, sorts
What EXPLAIN Shows  If WHERE, DISTINCT are used  Possible and actual indexes used  Length of index used  Approx # of records examined
Metadata  The MySQL optimizer uses metadata about cardinality, # rows, etc.  InnoDB has approximate statistics  InnoDB has one method of doing dives into the data  MyISAM has better and more accurate metadata
EXPLAIN Output EXPLAIN returns 10 fields: mysql> EXPLAIN SELECT return_date -> FROM rental WHERE rental_id = 13534G ******************* 1. row ******************* id: 1 select_type: SIMPLE table: rental type: const possible_keys: PRIMARY key: PRIMARY key_len: 4 ref: const rows: 1 Extra: 1 row in set (0.00 sec)
Id mysql> EXPLAIN SELECT return_date -> FROM rental WHERE rental_id = 13534G ******************* 1. row ******************* id: 1  Id = sequential identifier  One per table, subquery, derived table  No row returned for a view − Because it is virtual − Underlying tables are represented
select_type mysql> EXPLAIN SELECT return_date -> FROM rental WHERE rental_id = 13534G ******************* 1. row ******************* id: 1 select_type: SIMPLE  SIMPLE – one table, or JOINs  PRIMARY − First SELECT in a UNION − Outer query of a subquery  UNION, UNION RESULT
select_type in UNION queries mysql> EXPLAIN SELECT first_name FROM staff UNION SELECT first_name FROM customerG ********** 1. row ********** id: 1 key: NULL select_type: PRIMARY key_len: NULL table: staff ref: NULL type: ALL rows: 541 possible_keys: NULL Extra: key: NULL ********** 3. row ********** key_len: NULL id: NULL ref: NULL select_type: UNION RESULT rows: 1 table: <union1,2> Extra: type: ALL ********** 2. row ********** possible_keys: NULL id: 2 key: NULL select_type: UNION key_len: NULL table: customer ref: NULL type: ALL rows: NULL possible_keys: NULL Extra: key: NULL 3 rows in set (0.00 sec)
Other select_type output  Used in subqueries − More on subqueries later DEPENDENT UNION DEPENDENT SUBQUERY DERIVED UNCACHEABLE SUBQUERY
table mysql> EXPLAIN SELECT return_date -> FROM rental WHERE rental_id = 13534G ******************* 1. row ******************* id: 1 select_type: SIMPLE table: rental  Table name or alias used − Aliases like t1 are difficult to follow in long EXPLAIN plans  NULL table if no table is referenced or query is impossible
NULL table EXPLAIN SELECT 1+2G EXPLAIN SELECT return_date FROM rental WHERE rental_id=0G
type mysql> EXPLAIN SELECT return_date -> FROM rental WHERE rental_id = 13534G ******************* 1. row ******************* id: 1 select_type: SIMPLE table: rental type: const  “Data access method”  Get this as good as possible
type  ALL = full table scan − Everything else uses an index  index = full index scan − If you have to scan the entire data set, use a covering index to use a full index scan instead of a full table scan.  range = partial index scan − <, <=, >, >= − IS NULL, BETWEEN, IN
Range query EXPLAIN SELECT rental_id FROM rental WHERE rental_date BETWEEN '2006-02-14 00:00:00' and '2006-02-14 23:59:59'G  Not a range query; why? EXPLAIN SELECT rental_id FROM rental WHERE DATE(rental_date)='2006-02-14'G
type  index_subquery − subquery using a non-unique index of one table  unique subquery − Subquery using a PRIMARY or UNIQUE KEY of one table  More about subqueries later
type  index_merge − Use more than one index − Extra field shows more information  sort_union  intersection  union
Index Merge mysql> EXPLAIN SELECT customer_id FROM customer WHERE last_name LIKE “Hill%” OR customer_id<10G ******************* 1. row ******************* id: 1 select_type: SIMPLE table: customer type: index_merge possible_keys: PRIMARY,idx_last_name key: idx_last_name,PRIMARY key_len: 137,2 ref: NULL rows: 10 Extra: Using sort_union(idx_last_name, PRIMARY); Using where 1 row in set (0.03 sec)
ref_or_null  Joining/looking up non-unique index values  JOIN uses a non-unique index or key prefix  Indexed fields compared with = != <=>  Extra pass for NULL values if one may show up in the result
fulltext Data Access Strategy EXPLAIN SELECT film_id, title FROM film_text WHERE MATCH (title,description) AGAINST ('storm')G ******************** 1. row ******************** id: 1 select_type: SIMPLE table: film_text type: fulltext possible_keys: idx_title_description key: idx_title_description key_len: 0 ref: rows: 1 Extra: Using where 1 row in set (0.00 sec)
Non-unique index values EXPLAIN SELECT rental_id FROM rental WHERE customer_id=75G ******************** 1. row ******************** id: 1 select_type: SIMPLE table: rental type: ref possible_keys: idx_fk_customer_id key: idx_fk_customer_id key_len: 2 ref: const rows: 40 Extra: Using index 1 row in set (0.00 sec)  Like ref_or_null without the extra pass
ref  Joining/looking up non-unique index values  JOIN uses a non-unique index or key prefix  Indexed fields compared with = != <=>  Best data access strategy for non-unique values
eq_ref  Joining/looking up unique index values  JOIN uses a unique index or key prefix  Indexed fields compared with =
eq_ref Data Access Strategy mysql> EXPLAIN SELECT return_date -> FROM rental WHERE rental_id = 13534G ******************* 1. row ******************* id: 1 select_type: SIMPLE table: rental type: const possible_keys: PRIMARY key: PRIMARY key_len: 4 ref: const rows: 1 Extra: 1 row in set (0.00 sec)  Notes
eq_ref Data Access Strategy mysql> EXPLAIN SELECT first_name,last_name FROM rental -> INNER JOIN customer USING (customer_id) -> WHERE rental_date BETWEEN '2006-02-14 00:00:00' -> AND '2006-02-14 23:59:59'G ********** 1. row ********** ********** 2. row ********** id: 1 id: 1 select_type: SIMPLE select_type: SIMPLE table: rental table: customer type: range type: eq_ref possible_keys: possible_keys: PRIMARY rental_date,idx_fk_customer_i key: PRIMARY d key_len: 2 key: rental_date ref: key_len: 8 sakila.rental.customer_id ref: NULL rows: 1 rows: 2614 Extra: Extra: Using where; 2 rows in set (0.03 sec) Using index
Fastest Data Access strategies  Const – at most one value, uses PRIMARY or UNIQUE KEY mysql> EXPLAIN SELECT return_date FROM rental AS r WHERE rental_id = 13534G  System – system table, has 1 value EXPLAIN SELECT Time_zone_id, Use_leap_seconds FROM mysql.time_zoneG
Constant Propagation EXPLAIN SELECT return_date, first_name, last_name FROM rental INNER JOIN customer USING (customer_id) WHERE rental_id = 13534G ********** 1. row ********** id: 1 select_type: SIMPLE  The rental table uses table: rental rental_id as a filter, type: const possible_keys: and rental_id is a PRIMARY,idx_fk_customer_id key: PRIMARY unique field. key_len: 4  Thus, const makes ref: const rows: 1 sense Extra:  What about the customer table?
Constant Propagation EXPLAIN SELECT return_date, first_name, last_name FROM rental INNER JOIN customer USING (customer_id) WHERE rental_id = 13534G ********** 1. row ********** ********** 2. row ********** id: 1 id: 1 select_type: SIMPLE select_type: SIMPLE table: rental table: customer type: const type: const possible_keys: possible_keys: PRIMARY PRIMARY,idx_fk_customer_id key: PRIMARY key: PRIMARY key_len: 2 key_len: 4 ref: const ref: const rows: 1 rows: 1 Extra: Extra: 2 rows in set (0.00 sec)  Const is propagated because there is at most one customer_id, which is UNIQUE, NOT NULL
No data access strategy  No data access strategy when table is NULL  Fastest data access strategy − Because there is no strategy!  No data access strategy when WHERE is impossible − Optimizer only accesses metadata
EXPLAIN Plan indexes  possible_keys  key  key_len – longer keys take longer to look up and compare  ref – shows what is compared, field or “const”  Look closely if an index you think is possible is not considered
eq_ref Data Access Strategy mysql> EXPLAIN SELECT first_name,last_name FROM rental -> INNER JOIN customer USING (customer_id) -> WHERE rental_date BETWEEN '2006-02-14 00:00:00' -> AND '2006-02-14 23:59:59'G ********** 1. row ********** ********** 2. row ********** id: 1 id: 1 select_type: SIMPLE select_type: SIMPLE table: rental table: customer type: range type: eq_ref possible_keys: possible_keys: PRIMARY rental_date,idx_fk_customer_i key: PRIMARY d key_len: 2 key: rental_date ref: key_len: 8 sakila.rental.customer_id ref: NULL rows: 1 rows: 2614 Extra: Extra: Using where; 2 rows in set (0.03 sec) Using index
ref is “const” EXPLAIN SELECT return_date FROM rental WHERE rental_id = 13534G *************************** 1. row *************************** id: 1 select_type: SIMPLE table: rental type: const possible_keys: PRIMARY key: PRIMARY key_len: 4 ref: const rows: 1 Extra: 1 row in set (0.09 sec)
Approx # rows examined mysql> EXPLAIN SELECT return_date -> FROM rental WHERE rental_id = 13534G ******************* 1. row ******************* id: 1 select_type: SIMPLE table: rental type: const possible_keys: PRIMARY key: PRIMARY key_len: 4 ref: const rows: 1 Extra: 1 row in set (0.00 sec)
Approx # rows examined mysql> EXPLAIN SELECT first_name,last_name FROM customer LIMIT 10G *************** 1. row ***************** id: 1 select_type: SIMPLE table: customer type: ALL possible_keys: NULL key: NULL key_len: NULL ref: NULL rows: 541 Extra: 1 row in set (0.00 sec)  LIMIT does not change Rows, even though it affects # rows examined.
Extra  Can be good, bad, neutral − Sometimes you cannot avoid the bad  Distinct – stops after first row match  Full scan on NULL key – subquery lookup with no index (bad)  Impossible WHERE noticed after reading const tables  No tables used
Extra  Not exists – stops after first row match for each row set from previous tables  Select tables optimized away – Aggregate functions resolved by index or metadata (good)  Range checked for each record (index map: N) − No good index; may be one after values from previous tables are known
Extra: Using (...)  Extra: Using filesort – does an extra pass to sort the data. − Worse than using an index for sort order.  Index – uses index only, no table read − Covering index, good  Index for group-by − GROUP BY or DISTINCT resolved by index or metadata (good)  Temporary − Intermediate temporary table used (bad)
Extra & INFORMATION_SCHEMA  Scanned N databases − N is 0, 1 or all  Skip_open_table − Fastest, no table files need to be opened  Open_frm_only − Open the .frm file only  Open_trigger_only  Open_full_table − Open all the table files; slowest, can crash large systems
Sample subquery EXPLAIN mysql> EXPLAIN SELECT first_name,last_name,email -> IN (SELECT customer_id FROM rental AS rental_subquery WHERE return_date IS NULL) -> FROM customer AS customer_outerG ********* 1. row ******** ********** 2. row ********** id: 1 id: 2 select_type: PRIMARY select_type: DEPENDENT SUBQUERY table: customer_outer table: rental_subquery type: ALL type: index_subquery possible_keys: NULL possible_keys: idx_fk_customer_id key: NULL key: idx_fk_customer_id key_len: NULL key_len: 2 ref: NULL ref: func rows: 541 rows: 13 Extra: Extra: Using where; Full scan on NULL key 2 rows in set (0.00 sec)
MySQL and Subqueries  Avoid unoptimized subqueries − Not all subqueries, though that was true in earlier versions  Derived tables may be turned into views or intermediate temporary tables  Subqueries can be turned into joins in some cases  Getting better all the time − Used to be that all subqueries were dependent subqueries.
More EXPLAIN Values  EXPLAIN PARTITIONS − Adds a “partitions” value that shows a list of partitions to be checked for a partitioned table − NULL for a non-partitioned table  EXPLAIN EXTENDED − Adds filtered field, an approx % of how many of the examined rows will be returned − Show the query after the optimizer is finished with SHOW WARNINGS
EXPLAIN EXTENDED mysql> EXPLAIN EXTENDED SELECT customer_id -> FROM rental -> WHERE staff_id=2 AND inventory_id<100G ***************** 1. row ***************** id: 1 select_type: SIMPLE table: rental type: range possible_keys: idx_fk_inventory_id,idx_fk_staff_id key: idx_fk_inventory_id key_len: 3 ref: NULL rows: 326 filtered: 75.15 Extra: Using where 1 row in set, 1 warning (0.00 sec)
EXPLAIN EXTENDED mysql> SHOW WARNINGSG ***************** 1. row ***************** Level: Note Code: 1003 Message: select `sakila`.`rental`.`customer_id` AS `customer_id` from `sakila`.`rental` where ((`sakila`.`rental`.`staff_id` = 2) and (`sakila`.`rental`.`inventory_id` < 100)) 1 row in set (0.00 sec)
More EXPLAIN Information http://www.pythian.com/news/wp-content/uploads/explain-diagram.pdf Pages 590 – 614 of the MySQL Administrator's Bible Sakila sample database: http://dev.mysql.com/doc/index-other.html
Questions, comments, feedback? Sheeri K. Cabral – MySQL DBA www.oursql.com – MySQL podcast scabral@mozilla.com @sheeri – send me questions, I may be able to help! Worked with MySQL since 2001 Full-time MySQL DBA as of 2005
Optimizing Queries with EXPLAIN http://bit.ly/LTsbuJ Sheeri Cabral Senior DB Admin/Architect, Mozilla @sheeri www.sheeri.com
EXPLAIN  SQL extension  SELECT only  Can modify other statements: UPDATE tbl SET fld1=“foo” WHERE fld2=“bar”; can be changed to: EXPLAIN SELECT fld1,fld2 FROM tbl;
What EXPLAIN Shows  How many tables  How tables are joined  How data is looked up  If there are subqueries, unions, sorts
What EXPLAIN Shows  If WHERE, DISTINCT are used  Possible and actual indexes used  Length of index used  Approx # of records examined
Metadata  The MySQL optimizer uses metadata about cardinality, # rows, etc.  InnoDB has approximate statistics  InnoDB has one method of doing dives into the data  MyISAM has better and more accurate metadata
EXPLAIN Output EXPLAIN returns 10 fields: mysql> EXPLAIN SELECT return_date -> FROM rental WHERE rental_id = 13534G ******************* 1. row ******************* id: 1 select_type: SIMPLE table: rental type: const possible_keys: PRIMARY key: PRIMARY key_len: 4 ref: const rows: 1 Extra: 1 row in set (0.00 sec) One row per table.
Id mysql> EXPLAIN SELECT return_date -> FROM rental WHERE rental_id = 13534G ******************* 1. row ******************* id: 1  Id = sequential identifier  One per table, subquery, derived table  No row returned for a view − Because it is virtual − Underlying tables are represented One row per table.
select_type mysql> EXPLAIN SELECT return_date -> FROM rental WHERE rental_id = 13534G ******************* 1. row ******************* id: 1 select_type: SIMPLE  SIMPLE – one table, or JOINs  PRIMARY − First SELECT in a UNION − Outer query of a subquery  UNION, UNION RESULT One row per table.
select_type in UNION queries mysql> EXPLAIN SELECT first_name FROM staff UNION SELECT first_name FROM customerG ********** 1. row ********** id: 1 key: NULL select_type: PRIMARY key_len: NULL table: staff ref: NULL type: ALL rows: 541 possible_keys: NULL Extra: key: NULL ********** 3. row ********** key_len: NULL id: NULL ref: NULL select_type: UNION RESULT rows: 1 table: <union1,2> Extra: type: ALL ********** 2. row ********** possible_keys: NULL id: 2 key: NULL select_type: UNION key_len: NULL table: customer ref: NULL type: ALL rows: NULL possible_keys: NULL Extra: key: NULL 3 rows in set (0.00 sec) One row per table.
Other select_type output  Used in subqueries − More on subqueries later DEPENDENT UNION DEPENDENT SUBQUERY DERIVED UNCACHEABLE SUBQUERY One row per table.
table mysql> EXPLAIN SELECT return_date -> FROM rental WHERE rental_id = 13534G ******************* 1. row ******************* id: 1 select_type: SIMPLE table: rental  Table name or alias used − Aliases like t1 are difficult to follow in long EXPLAIN plans  NULL table if no table is referenced or query is impossible One row per table.
NULL table EXPLAIN SELECT 1+2G EXPLAIN SELECT return_date FROM rental WHERE rental_id=0G One row per table.
type mysql> EXPLAIN SELECT return_date -> FROM rental WHERE rental_id = 13534G ******************* 1. row ******************* id: 1 select_type: SIMPLE table: rental type: const  “Data access method”  Get this as good as possible One row per table.
type  ALL = full table scan − Everything else uses an index  index = full index scan − If you have to scan the entire data set, use a covering index to use a full index scan instead of a full table scan.  range = partial index scan − <, <=, >, >= − IS NULL, BETWEEN, IN One row per table.
Range query EXPLAIN SELECT rental_id FROM rental WHERE rental_date BETWEEN '2006-02-14 00:00:00' and '2006-02-14 23:59:59'G  Not a range query; why? EXPLAIN SELECT rental_id FROM rental WHERE DATE(rental_date)='2006-02-14'G One row per table.
type  index_subquery − subquery using a non-unique index of one table  unique subquery − Subquery using a PRIMARY or UNIQUE KEY of one table  More about subqueries later One row per table.
type  index_merge − Use more than one index − Extra field shows more information  sort_union  intersection  union One row per table.
Index Merge mysql> EXPLAIN SELECT customer_id FROM customer WHERE last_name LIKE “Hill%” OR customer_id<10G ******************* 1. row ******************* id: 1 select_type: SIMPLE table: customer type: index_merge possible_keys: PRIMARY,idx_last_name key: idx_last_name,PRIMARY key_len: 137,2 ref: NULL rows: 10 Extra: Using sort_union(idx_last_name, PRIMARY); Using where 1 row in set (0.03 sec) One row per table.
ref_or_null  Joining/looking up non-unique index values  JOIN uses a non-unique index or key prefix  Indexed fields compared with = != <=>  Extra pass for NULL values if one may show up in the result One row per table.
fulltext Data Access Strategy EXPLAIN SELECT film_id, title FROM film_text WHERE MATCH (title,description) AGAINST ('storm')G ******************** 1. row ******************** id: 1 select_type: SIMPLE table: film_text type: fulltext possible_keys: idx_title_description key: idx_title_description key_len: 0 ref: rows: 1 Extra: Using where 1 row in set (0.00 sec) One row per table.
Non-unique index values EXPLAIN SELECT rental_id FROM rental WHERE customer_id=75G ******************** 1. row ******************** id: 1 select_type: SIMPLE table: rental type: ref possible_keys: idx_fk_customer_id key: idx_fk_customer_id key_len: 2 ref: const rows: 40 Extra: Using index 1 row in set (0.00 sec)  Like ref_or_null without the extra pass One row per table.
ref  Joining/looking up non-unique index values  JOIN uses a non-unique index or key prefix  Indexed fields compared with = != <=>  Best data access strategy for non-unique values One row per table.
eq_ref  Joining/looking up unique index values  JOIN uses a unique index or key prefix  Indexed fields compared with = One row per table.
eq_ref Data Access Strategy mysql> EXPLAIN SELECT return_date -> FROM rental WHERE rental_id = 13534G ******************* 1. row ******************* id: 1 select_type: SIMPLE table: rental type: const possible_keys: PRIMARY key: PRIMARY key_len: 4 ref: const rows: 1 Extra: 1 row in set (0.00 sec)  Notes One row per table.
eq_ref Data Access Strategy mysql> EXPLAIN SELECT first_name,last_name FROM rental -> INNER JOIN customer USING (customer_id) -> WHERE rental_date BETWEEN '2006-02-14 00:00:00' -> AND '2006-02-14 23:59:59'G ********** 1. row ********** ********** 2. row ********** id: 1 id: 1 select_type: SIMPLE select_type: SIMPLE table: rental table: customer type: range type: eq_ref possible_keys: possible_keys: PRIMARY rental_date,idx_fk_customer_i key: PRIMARY d key_len: 2 key: rental_date ref: key_len: 8 sakila.rental.customer_id ref: NULL rows: 1 rows: 2614 Extra: Extra: Using where; 2 rows in set (0.03 sec) Using index One row per table.
Fastest Data Access strategies  Const – at most one value, uses PRIMARY or UNIQUE KEY mysql> EXPLAIN SELECT return_date FROM rental AS r WHERE rental_id = 13534G  System – system table, has 1 value EXPLAIN SELECT Time_zone_id, Use_leap_seconds FROM mysql.time_zoneG One row per table.
Constant Propagation EXPLAIN SELECT return_date, first_name, last_name FROM rental INNER JOIN customer USING (customer_id) WHERE rental_id = 13534G ********** 1. row ********** id: 1 select_type: SIMPLE  The rental table uses table: rental rental_id as a filter, type: const possible_keys: and rental_id is a PRIMARY,idx_fk_customer_id key: PRIMARY unique field. key_len: 4  Thus, const makes ref: const rows: 1 sense Extra:  What about the customer table? One row per table.
Constant Propagation EXPLAIN SELECT return_date, first_name, last_name FROM rental INNER JOIN customer USING (customer_id) WHERE rental_id = 13534G ********** 1. row ********** ********** 2. row ********** id: 1 id: 1 select_type: SIMPLE select_type: SIMPLE table: rental table: customer type: const type: const possible_keys: possible_keys: PRIMARY PRIMARY,idx_fk_customer_id key: PRIMARY key: PRIMARY key_len: 2 key_len: 4 ref: const ref: const rows: 1 rows: 1 Extra: Extra: 2 rows in set (0.00 sec)  Const is propagated because there is at most one customer_id, which is UNIQUE, NOT NULL One row per table.
No data access strategy  No data access strategy when table is NULL  Fastest data access strategy − Because there is no strategy!  No data access strategy when WHERE is impossible − Optimizer only accesses metadata One row per table.
EXPLAIN Plan indexes  possible_keys  key  key_len – longer keys take longer to look up and compare  ref – shows what is compared, field or “const”  Look closely if an index you think is possible is not considered One row per table.
eq_ref Data Access Strategy mysql> EXPLAIN SELECT first_name,last_name FROM rental -> INNER JOIN customer USING (customer_id) -> WHERE rental_date BETWEEN '2006-02-14 00:00:00' -> AND '2006-02-14 23:59:59'G ********** 1. row ********** ********** 2. row ********** id: 1 id: 1 select_type: SIMPLE select_type: SIMPLE table: rental table: customer type: range type: eq_ref possible_keys: possible_keys: PRIMARY rental_date,idx_fk_customer_i key: PRIMARY d key_len: 2 key: rental_date ref: key_len: 8 sakila.rental.customer_id ref: NULL rows: 1 rows: 2614 Extra: Extra: Using where; 2 rows in set (0.03 sec) Using index One row per table.
ref is “const” EXPLAIN SELECT return_date FROM rental WHERE rental_id = 13534G *************************** 1. row *************************** id: 1 select_type: SIMPLE table: rental type: const possible_keys: PRIMARY key: PRIMARY key_len: 4 ref: const rows: 1 Extra: 1 row in set (0.09 sec) One row per table.
Approx # rows examined mysql> EXPLAIN SELECT return_date -> FROM rental WHERE rental_id = 13534G ******************* 1. row ******************* id: 1 select_type: SIMPLE table: rental type: const possible_keys: PRIMARY key: PRIMARY key_len: 4 ref: const rows: 1 Extra: 1 row in set (0.00 sec) One row per table.
Approx # rows examined mysql> EXPLAIN SELECT first_name,last_name FROM customer LIMIT 10G *************** 1. row ***************** id: 1 select_type: SIMPLE table: customer type: ALL possible_keys: NULL key: NULL key_len: NULL ref: NULL rows: 541 Extra: 1 row in set (0.00 sec)  LIMIT does not change Rows, even though it affects # rows examined. One row per table.
Extra  Can be good, bad, neutral − Sometimes you cannot avoid the bad  Distinct – stops after first row match  Full scan on NULL key – subquery lookup with no index (bad)  Impossible WHERE noticed after reading const tables  No tables used One row per table.
Extra  Not exists – stops after first row match for each row set from previous tables  Select tables optimized away – Aggregate functions resolved by index or metadata (good)  Range checked for each record (index map: N) − No good index; may be one after values from previous tables are known One row per table.
Extra: Using (...)  Extra: Using filesort – does an extra pass to sort the data. − Worse than using an index for sort order.  Index – uses index only, no table read − Covering index, good  Index for group-by − GROUP BY or DISTINCT resolved by index or metadata (good)  Temporary − Intermediate temporary table used (bad) One row per table.
Extra & INFORMATION_SCHEMA  Scanned N databases − N is 0, 1 or all  Skip_open_table − Fastest, no table files need to be opened  Open_frm_only − Open the .frm file only  Open_trigger_only  Open_full_table − Open all the table files; slowest, can crash large systems One row per table.
Sample subquery EXPLAIN mysql> EXPLAIN SELECT first_name,last_name,email -> IN (SELECT customer_id FROM rental AS rental_subquery WHERE return_date IS NULL) -> FROM customer AS customer_outerG ********* 1. row ******** ********** 2. row ********** id: 1 id: 2 select_type: PRIMARY select_type: DEPENDENT SUBQUERY table: customer_outer table: rental_subquery type: ALL type: index_subquery possible_keys: NULL possible_keys: idx_fk_customer_id key: NULL key: idx_fk_customer_id key_len: NULL key_len: 2 ref: NULL ref: func rows: 541 rows: 13 Extra: Extra: Using where; Full scan on NULL key 2 rows in set (0.00 sec) One row per table.
MySQL and Subqueries  Avoid unoptimized subqueries − Not all subqueries, though that was true in earlier versions  Derived tables may be turned into views or intermediate temporary tables  Subqueries can be turned into joins in some cases  Getting better all the time − Used to be that all subqueries were dependent subqueries. One row per table.
More EXPLAIN Values  EXPLAIN PARTITIONS − Adds a “partitions” value that shows a list of partitions to be checked for a partitioned table − NULL for a non-partitioned table  EXPLAIN EXTENDED − Adds filtered field, an approx % of how many of the examined rows will be returned − Show the query after the optimizer is finished with SHOW WARNINGS One row per table.
EXPLAIN EXTENDED mysql> EXPLAIN EXTENDED SELECT customer_id -> FROM rental -> WHERE staff_id=2 AND inventory_id<100G ***************** 1. row ***************** id: 1 select_type: SIMPLE table: rental type: range possible_keys: idx_fk_inventory_id,idx_fk_staff_id key: idx_fk_inventory_id key_len: 3 ref: NULL rows: 326 filtered: 75.15 Extra: Using where 1 row in set, 1 warning (0.00 sec) Approx 75% of 326 rows will be returned. 326 rows = count(*) where inventory_id<100
EXPLAIN EXTENDED mysql> SHOW WARNINGSG ***************** 1. row ***************** Level: Note Code: 1003 Message: select `sakila`.`rental`.`customer_id` AS `customer_id` from `sakila`.`rental` where ((`sakila`.`rental`.`staff_id` = 2) and (`sakila`.`rental`.`inventory_id` < 100)) 1 row in set (0.00 sec) Approx 75% of 326 rows will be returned. 326 rows = count(*) where inventory_id<100
More EXPLAIN Information http://www.pythian.com/news/wp-content/uploads/explain-diagram.pdf Pages 590 – 614 of the MySQL Administrator's Bible Sakila sample database: http://dev.mysql.com/doc/index-other.html
Questions, comments, feedback? Sheeri K. Cabral – MySQL DBA www.oursql.com – MySQL podcast scabral@mozilla.com @sheeri – send me questions, I may be able to help! Worked with MySQL since 2001 Full-time MySQL DBA as of 2005

More Related Content

ODP
Explain
Ligaya Turmelle
 
PDF
56 Query Optimization
MYXPLAIN
 
PDF
Understanding Query Execution
webhostingguy
 
PDF
Mysql query optimization
Baohua Cai
 
PDF
Explaining the MySQL Explain
MYXPLAIN
 
PDF
0888 learning-mysql
sabir18
 
PDF
Advanced MySQL Query and Schema Tuning
MYXPLAIN
 
PDF
MySQL Query tuning 101
Sveta Smirnova
 
Explain
Ligaya Turmelle
 
56 Query Optimization
MYXPLAIN
 
Understanding Query Execution
webhostingguy
 
Mysql query optimization
Baohua Cai
 
Explaining the MySQL Explain
MYXPLAIN
 
0888 learning-mysql
sabir18
 
Advanced MySQL Query and Schema Tuning
MYXPLAIN
 
MySQL Query tuning 101
Sveta Smirnova
 

What's hot (18)

PDF
Advanced MySQL Query Tuning
Alexander Rubin
 
PDF
Troubleshooting MySQL Performance add-ons
Sveta Smirnova
 
PPTX
Optimizing queries MySQL
Georgi Sotirov
 
PDF
New features in Performance Schema 5.7 in action
Sveta Smirnova
 
PDF
Mysql Explain Explained
Jeremy Coates
 
PDF
Efficient Pagination Using MySQL
Evan Weaver
 
PDF
Load Data Fast!
Karwin Software Solutions LLC
 
PDF
Introduction into MySQL Query Tuning for Dev[Op]s
Sveta Smirnova
 
PDF
MySQL partitions tutorial
Giuseppe Maxia
 
PDF
Optimizer Histograms: When they Help and When Do Not?
Sveta Smirnova
 
PDF
Percona toolkit
Karwin Software Solutions LLC
 
PPT
Intro to my sql
Minhaj Kazi
 
PDF
Powerful Explain in MySQL 5.6
MYXPLAIN
 
PDF
介绍 MySQL
YUCHENG HU
 
PDF
The MySQL Query Optimizer Explained Through Optimizer Trace
oysteing
 
PDF
How to Use JSON in MySQL Wrong
Karwin Software Solutions LLC
 
PPTX
MySQL constraints
Harish Gyanani
 
PPTX
What's New In MySQL 5.6
Abdul Manaf
 
Advanced MySQL Query Tuning
Alexander Rubin
 
Troubleshooting MySQL Performance add-ons
Sveta Smirnova
 
Optimizing queries MySQL
Georgi Sotirov
 
New features in Performance Schema 5.7 in action
Sveta Smirnova
 
Mysql Explain Explained
Jeremy Coates
 
Efficient Pagination Using MySQL
Evan Weaver
 
Load Data Fast!
Karwin Software Solutions LLC
 
Introduction into MySQL Query Tuning for Dev[Op]s
Sveta Smirnova
 
MySQL partitions tutorial
Giuseppe Maxia
 
Optimizer Histograms: When they Help and When Do Not?
Sveta Smirnova
 
Percona toolkit
Karwin Software Solutions LLC
 
Intro to my sql
Minhaj Kazi
 
Powerful Explain in MySQL 5.6
MYXPLAIN
 
介绍 MySQL
YUCHENG HU
 
The MySQL Query Optimizer Explained Through Optimizer Trace
oysteing
 
How to Use JSON in MySQL Wrong
Karwin Software Solutions LLC
 
MySQL constraints
Harish Gyanani
 
What's New In MySQL 5.6
Abdul Manaf
 
Ad

Similar to Optimizing Queries with Explain (20)

PDF
Understanding query-execution806
yubao fu
 
PDF
MySQL Query And Index Tuning
Manikanda kumar
 
PPTX
Optimizing MySQL Queries
Achievers Tech
 
PDF
Explain2
Anis Berejeb
 
PDF
Covering indexes
MYXPLAIN
 
PPT
Explain that explain
Fabrizio Parrella
 
PDF
My sql explain cheat sheet
Achievers Tech
 
PDF
Scaling MySQL Strategies for Developers
Jonathan Levin
 
PDF
Advanced query optimization
MYXPLAIN
 
PDF
dbms.pdf
walter brand
 
PDF
Introduction to Databases - query optimizations for MySQL
Márton Kodok
 
PDF
Sql wksht-6
Mukesh Tekwani
 
PDF
Goldilocks and the Three MySQL Queries
Dave Stokes
 
PDF
Green plum培训材料
锐 张
 
PDF
Introduction into MySQL Query Tuning
Sveta Smirnova
 
PPTX
Inner join and outer join
Nargis Ehsan
 
ODP
MySQL Performance Optimization
Abhijit Mondal
 
PPTX
MYSQL join
Ahmed Farag
 
PDF
Bt0075 rdbms with mysql 2
Techglyphs
 
PDF
Non-Relational Postgres
EDB
 
Understanding query-execution806
yubao fu
 
MySQL Query And Index Tuning
Manikanda kumar
 
Optimizing MySQL Queries
Achievers Tech
 
Explain2
Anis Berejeb
 
Covering indexes
MYXPLAIN
 
Explain that explain
Fabrizio Parrella
 
My sql explain cheat sheet
Achievers Tech
 
Scaling MySQL Strategies for Developers
Jonathan Levin
 
Advanced query optimization
MYXPLAIN
 
dbms.pdf
walter brand
 
Introduction to Databases - query optimizations for MySQL
Márton Kodok
 
Sql wksht-6
Mukesh Tekwani
 
Goldilocks and the Three MySQL Queries
Dave Stokes
 
Green plum培训材料
锐 张
 
Introduction into MySQL Query Tuning
Sveta Smirnova
 
Inner join and outer join
Nargis Ehsan
 
MySQL Performance Optimization
Abhijit Mondal
 
MYSQL join
Ahmed Farag
 
Bt0075 rdbms with mysql 2
Techglyphs
 
Non-Relational Postgres
EDB
 
Ad

More from MYXPLAIN (12)

PDF
Query Optimization with MySQL 5.6: Old and New Tricks
MYXPLAIN
 
PDF
Need for Speed: MySQL Indexing
MYXPLAIN
 
PDF
Advanced Query Optimizer Tuning and Analysis
MYXPLAIN
 
PDF
MySQL Index Cookbook
MYXPLAIN
 
PDF
Are You Getting the Best of your MySQL Indexes
MYXPLAIN
 
PDF
How to Design Indexes, Really
MYXPLAIN
 
PDF
MySQL 5.6 Performance
MYXPLAIN
 
PPTX
MySQL Indexing - Best practices for MySQL 5.6
MYXPLAIN
 
PDF
Tools and Techniques for Index Design
MYXPLAIN
 
PDF
The Power of MySQL Explain
MYXPLAIN
 
PDF
Improving Performance with Better Indexes
MYXPLAIN
 
PDF
MySQL Optimizer Overview
MYXPLAIN
 
Query Optimization with MySQL 5.6: Old and New Tricks
MYXPLAIN
 
Need for Speed: MySQL Indexing
MYXPLAIN
 
Advanced Query Optimizer Tuning and Analysis
MYXPLAIN
 
MySQL Index Cookbook
MYXPLAIN
 
Are You Getting the Best of your MySQL Indexes
MYXPLAIN
 
How to Design Indexes, Really
MYXPLAIN
 
MySQL 5.6 Performance
MYXPLAIN
 
MySQL Indexing - Best practices for MySQL 5.6
MYXPLAIN
 
Tools and Techniques for Index Design
MYXPLAIN
 
The Power of MySQL Explain
MYXPLAIN
 
Improving Performance with Better Indexes
MYXPLAIN
 
MySQL Optimizer Overview
MYXPLAIN
 

Recently uploaded (20)

PDF
Mobile App Security Testing_ A Comprehensive Guide.pdf
flufftailshop
 
PDF
NewMind AI Weekly Chronicles - August'25-Week II
NewMind AI
 
PDF
Approach and Philosophy of On baking technology
30anthuong17
 
PDF
Build a system with the filesystem maintained by OSTree @ COSCUP 2025
Jian-Hong Pan
 
PDF
Per capita expenditure prediction using model stacking based on satellite ima...
IAESIJAI
 
PPTX
MYSQL Presentation for SQL database connectivity
Swati270511
 
PDF
Spectral efficient network and resource selection model in 5G networks
IAESIJAI
 
PDF
MIND Revenue Release Quarter 2 2025 Press Release
MIND CTI
 
PDF
cuic standard and advanced reporting.pdf
SimoneTitaniumTomase
 
PDF
The Rise and Fall of 3GPP – Time for a Sabbatical?
3G4G
 
PDF
Blue Purple Modern Animated Computer Science Presentation.pdf.pdf
Akar16
 
PPTX
KOM of Painting work and Equipment Insulation REV00 update 25-dec.pptx
muhammadmuneebnaeem7
 
PPT
Teaching material agriculture food technology
LiaRayya
 
PPTX
ACSFv1EN-58255 AWS Academy Cloud Security Foundations.pptx
23bcla24
 
PDF
A comparative analysis of optical character recognition models for extracting...
IAESIJAI
 
PDF
Dropbox Q2 2025 Financial Results & Investor Presentation
Dropbox
 
PPTX
Machine Learning_overview_presentation.pptx
kondavamsidharreddy2
 
PDF
Profit Center Accounting in SAP S/4HANA, S4F28 Col11
Libreria ERP
 
DOCX
The AUB Centre for AI in Media Proposal.docx
GAONE9
 
PPTX
A Presentation on Artificial Intelligence
memembruh336
 
Mobile App Security Testing_ A Comprehensive Guide.pdf
flufftailshop
 
NewMind AI Weekly Chronicles - August'25-Week II
NewMind AI
 
Approach and Philosophy of On baking technology
30anthuong17
 
Build a system with the filesystem maintained by OSTree @ COSCUP 2025
Jian-Hong Pan
 
Per capita expenditure prediction using model stacking based on satellite ima...
IAESIJAI
 
MYSQL Presentation for SQL database connectivity
Swati270511
 
Spectral efficient network and resource selection model in 5G networks
IAESIJAI
 
MIND Revenue Release Quarter 2 2025 Press Release
MIND CTI
 
cuic standard and advanced reporting.pdf
SimoneTitaniumTomase
 
The Rise and Fall of 3GPP – Time for a Sabbatical?
3G4G
 
Blue Purple Modern Animated Computer Science Presentation.pdf.pdf
Akar16
 
KOM of Painting work and Equipment Insulation REV00 update 25-dec.pptx
muhammadmuneebnaeem7
 
Teaching material agriculture food technology
LiaRayya
 
ACSFv1EN-58255 AWS Academy Cloud Security Foundations.pptx
23bcla24
 
A comparative analysis of optical character recognition models for extracting...
IAESIJAI
 
Dropbox Q2 2025 Financial Results & Investor Presentation
Dropbox
 
Machine Learning_overview_presentation.pptx
kondavamsidharreddy2
 
Profit Center Accounting in SAP S/4HANA, S4F28 Col11
Libreria ERP
 
The AUB Centre for AI in Media Proposal.docx
GAONE9
 
A Presentation on Artificial Intelligence
memembruh336
 

Optimizing Queries with Explain

  • 1. Optimizing Queries with EXPLAIN http://bit.ly/LTsbuJ Sheeri Cabral Senior DB Admin/Architect, Mozilla @sheeri www.sheeri.com
  • 2. EXPLAIN  SQL extension  SELECT only  Can modify other statements: UPDATE tbl SET fld1=“foo” WHERE fld2=“bar”; can be changed to: EXPLAIN SELECT fld1,fld2 FROM tbl;
  • 3. What EXPLAIN Shows  How many tables  How tables are joined  How data is looked up  If there are subqueries, unions, sorts
  • 4. What EXPLAIN Shows  If WHERE, DISTINCT are used  Possible and actual indexes used  Length of index used  Approx # of records examined
  • 5. Metadata  The MySQL optimizer uses metadata about cardinality, # rows, etc.  InnoDB has approximate statistics  InnoDB has one method of doing dives into the data  MyISAM has better and more accurate metadata
  • 6. EXPLAIN Output EXPLAIN returns 10 fields: mysql> EXPLAIN SELECT return_date -> FROM rental WHERE rental_id = 13534G ******************* 1. row ******************* id: 1 select_type: SIMPLE table: rental type: const possible_keys: PRIMARY key: PRIMARY key_len: 4 ref: const rows: 1 Extra: 1 row in set (0.00 sec)
  • 7. Id mysql> EXPLAIN SELECT return_date -> FROM rental WHERE rental_id = 13534G ******************* 1. row ******************* id: 1  Id = sequential identifier  One per table, subquery, derived table  No row returned for a view − Because it is virtual − Underlying tables are represented
  • 8. select_type mysql> EXPLAIN SELECT return_date -> FROM rental WHERE rental_id = 13534G ******************* 1. row ******************* id: 1 select_type: SIMPLE  SIMPLE – one table, or JOINs  PRIMARY − First SELECT in a UNION − Outer query of a subquery  UNION, UNION RESULT
  • 9. select_type in UNION queries mysql> EXPLAIN SELECT first_name FROM staff UNION SELECT first_name FROM customerG ********** 1. row ********** id: 1 key: NULL select_type: PRIMARY key_len: NULL table: staff ref: NULL type: ALL rows: 541 possible_keys: NULL Extra: key: NULL ********** 3. row ********** key_len: NULL id: NULL ref: NULL select_type: UNION RESULT rows: 1 table: <union1,2> Extra: type: ALL ********** 2. row ********** possible_keys: NULL id: 2 key: NULL select_type: UNION key_len: NULL table: customer ref: NULL type: ALL rows: NULL possible_keys: NULL Extra: key: NULL 3 rows in set (0.00 sec)
  • 10. Other select_type output  Used in subqueries − More on subqueries later DEPENDENT UNION DEPENDENT SUBQUERY DERIVED UNCACHEABLE SUBQUERY
  • 11. table mysql> EXPLAIN SELECT return_date -> FROM rental WHERE rental_id = 13534G ******************* 1. row ******************* id: 1 select_type: SIMPLE table: rental  Table name or alias used − Aliases like t1 are difficult to follow in long EXPLAIN plans  NULL table if no table is referenced or query is impossible
  • 12. NULL table EXPLAIN SELECT 1+2G EXPLAIN SELECT return_date FROM rental WHERE rental_id=0G
  • 13. type mysql> EXPLAIN SELECT return_date -> FROM rental WHERE rental_id = 13534G ******************* 1. row ******************* id: 1 select_type: SIMPLE table: rental type: const  “Data access method”  Get this as good as possible
  • 14. type  ALL = full table scan − Everything else uses an index  index = full index scan − If you have to scan the entire data set, use a covering index to use a full index scan instead of a full table scan.  range = partial index scan − <, <=, >, >= − IS NULL, BETWEEN, IN
  • 15. Range query EXPLAIN SELECT rental_id FROM rental WHERE rental_date BETWEEN '2006-02-14 00:00:00' and '2006-02-14 23:59:59'G  Not a range query; why? EXPLAIN SELECT rental_id FROM rental WHERE DATE(rental_date)='2006-02-14'G
  • 16. type  index_subquery − subquery using a non-unique index of one table  unique subquery − Subquery using a PRIMARY or UNIQUE KEY of one table  More about subqueries later
  • 17. type  index_merge − Use more than one index − Extra field shows more information  sort_union  intersection  union
  • 18. Index Merge mysql> EXPLAIN SELECT customer_id FROM customer WHERE last_name LIKE “Hill%” OR customer_id<10G ******************* 1. row ******************* id: 1 select_type: SIMPLE table: customer type: index_merge possible_keys: PRIMARY,idx_last_name key: idx_last_name,PRIMARY key_len: 137,2 ref: NULL rows: 10 Extra: Using sort_union(idx_last_name, PRIMARY); Using where 1 row in set (0.03 sec)
  • 19. ref_or_null  Joining/looking up non-unique index values  JOIN uses a non-unique index or key prefix  Indexed fields compared with = != <=>  Extra pass for NULL values if one may show up in the result
  • 20. fulltext Data Access Strategy EXPLAIN SELECT film_id, title FROM film_text WHERE MATCH (title,description) AGAINST ('storm')G ******************** 1. row ******************** id: 1 select_type: SIMPLE table: film_text type: fulltext possible_keys: idx_title_description key: idx_title_description key_len: 0 ref: rows: 1 Extra: Using where 1 row in set (0.00 sec)
  • 21. Non-unique index values EXPLAIN SELECT rental_id FROM rental WHERE customer_id=75G ******************** 1. row ******************** id: 1 select_type: SIMPLE table: rental type: ref possible_keys: idx_fk_customer_id key: idx_fk_customer_id key_len: 2 ref: const rows: 40 Extra: Using index 1 row in set (0.00 sec)  Like ref_or_null without the extra pass
  • 22. ref  Joining/looking up non-unique index values  JOIN uses a non-unique index or key prefix  Indexed fields compared with = != <=>  Best data access strategy for non-unique values
  • 23. eq_ref  Joining/looking up unique index values  JOIN uses a unique index or key prefix  Indexed fields compared with =
  • 24. eq_ref Data Access Strategy mysql> EXPLAIN SELECT return_date -> FROM rental WHERE rental_id = 13534G ******************* 1. row ******************* id: 1 select_type: SIMPLE table: rental type: const possible_keys: PRIMARY key: PRIMARY key_len: 4 ref: const rows: 1 Extra: 1 row in set (0.00 sec)  Notes
  • 25. eq_ref Data Access Strategy mysql> EXPLAIN SELECT first_name,last_name FROM rental -> INNER JOIN customer USING (customer_id) -> WHERE rental_date BETWEEN '2006-02-14 00:00:00' -> AND '2006-02-14 23:59:59'G ********** 1. row ********** ********** 2. row ********** id: 1 id: 1 select_type: SIMPLE select_type: SIMPLE table: rental table: customer type: range type: eq_ref possible_keys: possible_keys: PRIMARY rental_date,idx_fk_customer_i key: PRIMARY d key_len: 2 key: rental_date ref: key_len: 8 sakila.rental.customer_id ref: NULL rows: 1 rows: 2614 Extra: Extra: Using where; 2 rows in set (0.03 sec) Using index
  • 26. Fastest Data Access strategies  Const – at most one value, uses PRIMARY or UNIQUE KEY mysql> EXPLAIN SELECT return_date FROM rental AS r WHERE rental_id = 13534G  System – system table, has 1 value EXPLAIN SELECT Time_zone_id, Use_leap_seconds FROM mysql.time_zoneG
  • 27. Constant Propagation EXPLAIN SELECT return_date, first_name, last_name FROM rental INNER JOIN customer USING (customer_id) WHERE rental_id = 13534G ********** 1. row ********** id: 1 select_type: SIMPLE  The rental table uses table: rental rental_id as a filter, type: const possible_keys: and rental_id is a PRIMARY,idx_fk_customer_id key: PRIMARY unique field. key_len: 4  Thus, const makes ref: const rows: 1 sense Extra:  What about the customer table?
  • 28. Constant Propagation EXPLAIN SELECT return_date, first_name, last_name FROM rental INNER JOIN customer USING (customer_id) WHERE rental_id = 13534G ********** 1. row ********** ********** 2. row ********** id: 1 id: 1 select_type: SIMPLE select_type: SIMPLE table: rental table: customer type: const type: const possible_keys: possible_keys: PRIMARY PRIMARY,idx_fk_customer_id key: PRIMARY key: PRIMARY key_len: 2 key_len: 4 ref: const ref: const rows: 1 rows: 1 Extra: Extra: 2 rows in set (0.00 sec)  Const is propagated because there is at most one customer_id, which is UNIQUE, NOT NULL
  • 29. No data access strategy  No data access strategy when table is NULL  Fastest data access strategy − Because there is no strategy!  No data access strategy when WHERE is impossible − Optimizer only accesses metadata
  • 30. EXPLAIN Plan indexes  possible_keys  key  key_len – longer keys take longer to look up and compare  ref – shows what is compared, field or “const”  Look closely if an index you think is possible is not considered
  • 31. eq_ref Data Access Strategy mysql> EXPLAIN SELECT first_name,last_name FROM rental -> INNER JOIN customer USING (customer_id) -> WHERE rental_date BETWEEN '2006-02-14 00:00:00' -> AND '2006-02-14 23:59:59'G ********** 1. row ********** ********** 2. row ********** id: 1 id: 1 select_type: SIMPLE select_type: SIMPLE table: rental table: customer type: range type: eq_ref possible_keys: possible_keys: PRIMARY rental_date,idx_fk_customer_i key: PRIMARY d key_len: 2 key: rental_date ref: key_len: 8 sakila.rental.customer_id ref: NULL rows: 1 rows: 2614 Extra: Extra: Using where; 2 rows in set (0.03 sec) Using index
  • 32. ref is “const” EXPLAIN SELECT return_date FROM rental WHERE rental_id = 13534G *************************** 1. row *************************** id: 1 select_type: SIMPLE table: rental type: const possible_keys: PRIMARY key: PRIMARY key_len: 4 ref: const rows: 1 Extra: 1 row in set (0.09 sec)
  • 33. Approx # rows examined mysql> EXPLAIN SELECT return_date -> FROM rental WHERE rental_id = 13534G ******************* 1. row ******************* id: 1 select_type: SIMPLE table: rental type: const possible_keys: PRIMARY key: PRIMARY key_len: 4 ref: const rows: 1 Extra: 1 row in set (0.00 sec)
  • 34. Approx # rows examined mysql> EXPLAIN SELECT first_name,last_name FROM customer LIMIT 10G *************** 1. row ***************** id: 1 select_type: SIMPLE table: customer type: ALL possible_keys: NULL key: NULL key_len: NULL ref: NULL rows: 541 Extra: 1 row in set (0.00 sec)  LIMIT does not change Rows, even though it affects # rows examined.
  • 35. Extra  Can be good, bad, neutral − Sometimes you cannot avoid the bad  Distinct – stops after first row match  Full scan on NULL key – subquery lookup with no index (bad)  Impossible WHERE noticed after reading const tables  No tables used
  • 36. Extra  Not exists – stops after first row match for each row set from previous tables  Select tables optimized away – Aggregate functions resolved by index or metadata (good)  Range checked for each record (index map: N) − No good index; may be one after values from previous tables are known
  • 37. Extra: Using (...)  Extra: Using filesort – does an extra pass to sort the data. − Worse than using an index for sort order.  Index – uses index only, no table read − Covering index, good  Index for group-by − GROUP BY or DISTINCT resolved by index or metadata (good)  Temporary − Intermediate temporary table used (bad)
  • 38. Extra & INFORMATION_SCHEMA  Scanned N databases − N is 0, 1 or all  Skip_open_table − Fastest, no table files need to be opened  Open_frm_only − Open the .frm file only  Open_trigger_only  Open_full_table − Open all the table files; slowest, can crash large systems
  • 39. Sample subquery EXPLAIN mysql> EXPLAIN SELECT first_name,last_name,email -> IN (SELECT customer_id FROM rental AS rental_subquery WHERE return_date IS NULL) -> FROM customer AS customer_outerG ********* 1. row ******** ********** 2. row ********** id: 1 id: 2 select_type: PRIMARY select_type: DEPENDENT SUBQUERY table: customer_outer table: rental_subquery type: ALL type: index_subquery possible_keys: NULL possible_keys: idx_fk_customer_id key: NULL key: idx_fk_customer_id key_len: NULL key_len: 2 ref: NULL ref: func rows: 541 rows: 13 Extra: Extra: Using where; Full scan on NULL key 2 rows in set (0.00 sec)
  • 40. MySQL and Subqueries  Avoid unoptimized subqueries − Not all subqueries, though that was true in earlier versions  Derived tables may be turned into views or intermediate temporary tables  Subqueries can be turned into joins in some cases  Getting better all the time − Used to be that all subqueries were dependent subqueries.
  • 41. More EXPLAIN Values  EXPLAIN PARTITIONS − Adds a “partitions” value that shows a list of partitions to be checked for a partitioned table − NULL for a non-partitioned table  EXPLAIN EXTENDED − Adds filtered field, an approx % of how many of the examined rows will be returned − Show the query after the optimizer is finished with SHOW WARNINGS
  • 42. EXPLAIN EXTENDED mysql> EXPLAIN EXTENDED SELECT customer_id -> FROM rental -> WHERE staff_id=2 AND inventory_id<100G ***************** 1. row ***************** id: 1 select_type: SIMPLE table: rental type: range possible_keys: idx_fk_inventory_id,idx_fk_staff_id key: idx_fk_inventory_id key_len: 3 ref: NULL rows: 326 filtered: 75.15 Extra: Using where 1 row in set, 1 warning (0.00 sec)
  • 43. EXPLAIN EXTENDED mysql> SHOW WARNINGSG ***************** 1. row ***************** Level: Note Code: 1003 Message: select `sakila`.`rental`.`customer_id` AS `customer_id` from `sakila`.`rental` where ((`sakila`.`rental`.`staff_id` = 2) and (`sakila`.`rental`.`inventory_id` < 100)) 1 row in set (0.00 sec)
  • 44. More EXPLAIN Information http://www.pythian.com/news/wp-content/uploads/explain-diagram.pdf Pages 590 – 614 of the MySQL Administrator's Bible Sakila sample database: http://dev.mysql.com/doc/index-other.html
  • 45. Questions, comments, feedback? Sheeri K. Cabral – MySQL DBA www.oursql.com – MySQL podcast [email protected] @sheeri – send me questions, I may be able to help! Worked with MySQL since 2001 Full-time MySQL DBA as of 2005
  • 46. Optimizing Queries with EXPLAIN http://bit.ly/LTsbuJ Sheeri Cabral Senior DB Admin/Architect, Mozilla @sheeri www.sheeri.com
  • 47. EXPLAIN  SQL extension  SELECT only  Can modify other statements: UPDATE tbl SET fld1=“foo” WHERE fld2=“bar”; can be changed to: EXPLAIN SELECT fld1,fld2 FROM tbl;
  • 48. What EXPLAIN Shows  How many tables  How tables are joined  How data is looked up  If there are subqueries, unions, sorts
  • 49. What EXPLAIN Shows  If WHERE, DISTINCT are used  Possible and actual indexes used  Length of index used  Approx # of records examined
  • 50. Metadata  The MySQL optimizer uses metadata about cardinality, # rows, etc.  InnoDB has approximate statistics  InnoDB has one method of doing dives into the data  MyISAM has better and more accurate metadata
  • 51. EXPLAIN Output EXPLAIN returns 10 fields: mysql> EXPLAIN SELECT return_date -> FROM rental WHERE rental_id = 13534G ******************* 1. row ******************* id: 1 select_type: SIMPLE table: rental type: const possible_keys: PRIMARY key: PRIMARY key_len: 4 ref: const rows: 1 Extra: 1 row in set (0.00 sec) One row per table.
  • 52. Id mysql> EXPLAIN SELECT return_date -> FROM rental WHERE rental_id = 13534G ******************* 1. row ******************* id: 1  Id = sequential identifier  One per table, subquery, derived table  No row returned for a view − Because it is virtual − Underlying tables are represented One row per table.
  • 53. select_type mysql> EXPLAIN SELECT return_date -> FROM rental WHERE rental_id = 13534G ******************* 1. row ******************* id: 1 select_type: SIMPLE  SIMPLE – one table, or JOINs  PRIMARY − First SELECT in a UNION − Outer query of a subquery  UNION, UNION RESULT One row per table.
  • 54. select_type in UNION queries mysql> EXPLAIN SELECT first_name FROM staff UNION SELECT first_name FROM customerG ********** 1. row ********** id: 1 key: NULL select_type: PRIMARY key_len: NULL table: staff ref: NULL type: ALL rows: 541 possible_keys: NULL Extra: key: NULL ********** 3. row ********** key_len: NULL id: NULL ref: NULL select_type: UNION RESULT rows: 1 table: <union1,2> Extra: type: ALL ********** 2. row ********** possible_keys: NULL id: 2 key: NULL select_type: UNION key_len: NULL table: customer ref: NULL type: ALL rows: NULL possible_keys: NULL Extra: key: NULL 3 rows in set (0.00 sec) One row per table.
  • 55. Other select_type output  Used in subqueries − More on subqueries later DEPENDENT UNION DEPENDENT SUBQUERY DERIVED UNCACHEABLE SUBQUERY One row per table.
  • 56. table mysql> EXPLAIN SELECT return_date -> FROM rental WHERE rental_id = 13534G ******************* 1. row ******************* id: 1 select_type: SIMPLE table: rental  Table name or alias used − Aliases like t1 are difficult to follow in long EXPLAIN plans  NULL table if no table is referenced or query is impossible One row per table.
  • 57. NULL table EXPLAIN SELECT 1+2G EXPLAIN SELECT return_date FROM rental WHERE rental_id=0G One row per table.
  • 58. type mysql> EXPLAIN SELECT return_date -> FROM rental WHERE rental_id = 13534G ******************* 1. row ******************* id: 1 select_type: SIMPLE table: rental type: const  “Data access method”  Get this as good as possible One row per table.
  • 59. type  ALL = full table scan − Everything else uses an index  index = full index scan − If you have to scan the entire data set, use a covering index to use a full index scan instead of a full table scan.  range = partial index scan − <, <=, >, >= − IS NULL, BETWEEN, IN One row per table.
  • 60. Range query EXPLAIN SELECT rental_id FROM rental WHERE rental_date BETWEEN '2006-02-14 00:00:00' and '2006-02-14 23:59:59'G  Not a range query; why? EXPLAIN SELECT rental_id FROM rental WHERE DATE(rental_date)='2006-02-14'G One row per table.
  • 61. type  index_subquery − subquery using a non-unique index of one table  unique subquery − Subquery using a PRIMARY or UNIQUE KEY of one table  More about subqueries later One row per table.
  • 62. type  index_merge − Use more than one index − Extra field shows more information  sort_union  intersection  union One row per table.
  • 63. Index Merge mysql> EXPLAIN SELECT customer_id FROM customer WHERE last_name LIKE “Hill%” OR customer_id<10G ******************* 1. row ******************* id: 1 select_type: SIMPLE table: customer type: index_merge possible_keys: PRIMARY,idx_last_name key: idx_last_name,PRIMARY key_len: 137,2 ref: NULL rows: 10 Extra: Using sort_union(idx_last_name, PRIMARY); Using where 1 row in set (0.03 sec) One row per table.
  • 64. ref_or_null  Joining/looking up non-unique index values  JOIN uses a non-unique index or key prefix  Indexed fields compared with = != <=>  Extra pass for NULL values if one may show up in the result One row per table.
  • 65. fulltext Data Access Strategy EXPLAIN SELECT film_id, title FROM film_text WHERE MATCH (title,description) AGAINST ('storm')G ******************** 1. row ******************** id: 1 select_type: SIMPLE table: film_text type: fulltext possible_keys: idx_title_description key: idx_title_description key_len: 0 ref: rows: 1 Extra: Using where 1 row in set (0.00 sec) One row per table.
  • 66. Non-unique index values EXPLAIN SELECT rental_id FROM rental WHERE customer_id=75G ******************** 1. row ******************** id: 1 select_type: SIMPLE table: rental type: ref possible_keys: idx_fk_customer_id key: idx_fk_customer_id key_len: 2 ref: const rows: 40 Extra: Using index 1 row in set (0.00 sec)  Like ref_or_null without the extra pass One row per table.
  • 67. ref  Joining/looking up non-unique index values  JOIN uses a non-unique index or key prefix  Indexed fields compared with = != <=>  Best data access strategy for non-unique values One row per table.
  • 68. eq_ref  Joining/looking up unique index values  JOIN uses a unique index or key prefix  Indexed fields compared with = One row per table.
  • 69. eq_ref Data Access Strategy mysql> EXPLAIN SELECT return_date -> FROM rental WHERE rental_id = 13534G ******************* 1. row ******************* id: 1 select_type: SIMPLE table: rental type: const possible_keys: PRIMARY key: PRIMARY key_len: 4 ref: const rows: 1 Extra: 1 row in set (0.00 sec)  Notes One row per table.
  • 70. eq_ref Data Access Strategy mysql> EXPLAIN SELECT first_name,last_name FROM rental -> INNER JOIN customer USING (customer_id) -> WHERE rental_date BETWEEN '2006-02-14 00:00:00' -> AND '2006-02-14 23:59:59'G ********** 1. row ********** ********** 2. row ********** id: 1 id: 1 select_type: SIMPLE select_type: SIMPLE table: rental table: customer type: range type: eq_ref possible_keys: possible_keys: PRIMARY rental_date,idx_fk_customer_i key: PRIMARY d key_len: 2 key: rental_date ref: key_len: 8 sakila.rental.customer_id ref: NULL rows: 1 rows: 2614 Extra: Extra: Using where; 2 rows in set (0.03 sec) Using index One row per table.
  • 71. Fastest Data Access strategies  Const – at most one value, uses PRIMARY or UNIQUE KEY mysql> EXPLAIN SELECT return_date FROM rental AS r WHERE rental_id = 13534G  System – system table, has 1 value EXPLAIN SELECT Time_zone_id, Use_leap_seconds FROM mysql.time_zoneG One row per table.
  • 72. Constant Propagation EXPLAIN SELECT return_date, first_name, last_name FROM rental INNER JOIN customer USING (customer_id) WHERE rental_id = 13534G ********** 1. row ********** id: 1 select_type: SIMPLE  The rental table uses table: rental rental_id as a filter, type: const possible_keys: and rental_id is a PRIMARY,idx_fk_customer_id key: PRIMARY unique field. key_len: 4  Thus, const makes ref: const rows: 1 sense Extra:  What about the customer table? One row per table.
  • 73. Constant Propagation EXPLAIN SELECT return_date, first_name, last_name FROM rental INNER JOIN customer USING (customer_id) WHERE rental_id = 13534G ********** 1. row ********** ********** 2. row ********** id: 1 id: 1 select_type: SIMPLE select_type: SIMPLE table: rental table: customer type: const type: const possible_keys: possible_keys: PRIMARY PRIMARY,idx_fk_customer_id key: PRIMARY key: PRIMARY key_len: 2 key_len: 4 ref: const ref: const rows: 1 rows: 1 Extra: Extra: 2 rows in set (0.00 sec)  Const is propagated because there is at most one customer_id, which is UNIQUE, NOT NULL One row per table.
  • 74. No data access strategy  No data access strategy when table is NULL  Fastest data access strategy − Because there is no strategy!  No data access strategy when WHERE is impossible − Optimizer only accesses metadata One row per table.
  • 75. EXPLAIN Plan indexes  possible_keys  key  key_len – longer keys take longer to look up and compare  ref – shows what is compared, field or “const”  Look closely if an index you think is possible is not considered One row per table.
  • 76. eq_ref Data Access Strategy mysql> EXPLAIN SELECT first_name,last_name FROM rental -> INNER JOIN customer USING (customer_id) -> WHERE rental_date BETWEEN '2006-02-14 00:00:00' -> AND '2006-02-14 23:59:59'G ********** 1. row ********** ********** 2. row ********** id: 1 id: 1 select_type: SIMPLE select_type: SIMPLE table: rental table: customer type: range type: eq_ref possible_keys: possible_keys: PRIMARY rental_date,idx_fk_customer_i key: PRIMARY d key_len: 2 key: rental_date ref: key_len: 8 sakila.rental.customer_id ref: NULL rows: 1 rows: 2614 Extra: Extra: Using where; 2 rows in set (0.03 sec) Using index One row per table.
  • 77. ref is “const” EXPLAIN SELECT return_date FROM rental WHERE rental_id = 13534G *************************** 1. row *************************** id: 1 select_type: SIMPLE table: rental type: const possible_keys: PRIMARY key: PRIMARY key_len: 4 ref: const rows: 1 Extra: 1 row in set (0.09 sec) One row per table.
  • 78. Approx # rows examined mysql> EXPLAIN SELECT return_date -> FROM rental WHERE rental_id = 13534G ******************* 1. row ******************* id: 1 select_type: SIMPLE table: rental type: const possible_keys: PRIMARY key: PRIMARY key_len: 4 ref: const rows: 1 Extra: 1 row in set (0.00 sec) One row per table.
  • 79. Approx # rows examined mysql> EXPLAIN SELECT first_name,last_name FROM customer LIMIT 10G *************** 1. row ***************** id: 1 select_type: SIMPLE table: customer type: ALL possible_keys: NULL key: NULL key_len: NULL ref: NULL rows: 541 Extra: 1 row in set (0.00 sec)  LIMIT does not change Rows, even though it affects # rows examined. One row per table.
  • 80. Extra  Can be good, bad, neutral − Sometimes you cannot avoid the bad  Distinct – stops after first row match  Full scan on NULL key – subquery lookup with no index (bad)  Impossible WHERE noticed after reading const tables  No tables used One row per table.
  • 81. Extra  Not exists – stops after first row match for each row set from previous tables  Select tables optimized away – Aggregate functions resolved by index or metadata (good)  Range checked for each record (index map: N) − No good index; may be one after values from previous tables are known One row per table.
  • 82. Extra: Using (...)  Extra: Using filesort – does an extra pass to sort the data. − Worse than using an index for sort order.  Index – uses index only, no table read − Covering index, good  Index for group-by − GROUP BY or DISTINCT resolved by index or metadata (good)  Temporary − Intermediate temporary table used (bad) One row per table.
  • 83. Extra & INFORMATION_SCHEMA  Scanned N databases − N is 0, 1 or all  Skip_open_table − Fastest, no table files need to be opened  Open_frm_only − Open the .frm file only  Open_trigger_only  Open_full_table − Open all the table files; slowest, can crash large systems One row per table.
  • 84. Sample subquery EXPLAIN mysql> EXPLAIN SELECT first_name,last_name,email -> IN (SELECT customer_id FROM rental AS rental_subquery WHERE return_date IS NULL) -> FROM customer AS customer_outerG ********* 1. row ******** ********** 2. row ********** id: 1 id: 2 select_type: PRIMARY select_type: DEPENDENT SUBQUERY table: customer_outer table: rental_subquery type: ALL type: index_subquery possible_keys: NULL possible_keys: idx_fk_customer_id key: NULL key: idx_fk_customer_id key_len: NULL key_len: 2 ref: NULL ref: func rows: 541 rows: 13 Extra: Extra: Using where; Full scan on NULL key 2 rows in set (0.00 sec) One row per table.
  • 85. MySQL and Subqueries  Avoid unoptimized subqueries − Not all subqueries, though that was true in earlier versions  Derived tables may be turned into views or intermediate temporary tables  Subqueries can be turned into joins in some cases  Getting better all the time − Used to be that all subqueries were dependent subqueries. One row per table.
  • 86. More EXPLAIN Values  EXPLAIN PARTITIONS − Adds a “partitions” value that shows a list of partitions to be checked for a partitioned table − NULL for a non-partitioned table  EXPLAIN EXTENDED − Adds filtered field, an approx % of how many of the examined rows will be returned − Show the query after the optimizer is finished with SHOW WARNINGS One row per table.
  • 87. EXPLAIN EXTENDED mysql> EXPLAIN EXTENDED SELECT customer_id -> FROM rental -> WHERE staff_id=2 AND inventory_id<100G ***************** 1. row ***************** id: 1 select_type: SIMPLE table: rental type: range possible_keys: idx_fk_inventory_id,idx_fk_staff_id key: idx_fk_inventory_id key_len: 3 ref: NULL rows: 326 filtered: 75.15 Extra: Using where 1 row in set, 1 warning (0.00 sec) Approx 75% of 326 rows will be returned. 326 rows = count(*) where inventory_id<100
  • 88. EXPLAIN EXTENDED mysql> SHOW WARNINGSG ***************** 1. row ***************** Level: Note Code: 1003 Message: select `sakila`.`rental`.`customer_id` AS `customer_id` from `sakila`.`rental` where ((`sakila`.`rental`.`staff_id` = 2) and (`sakila`.`rental`.`inventory_id` < 100)) 1 row in set (0.00 sec) Approx 75% of 326 rows will be returned. 326 rows = count(*) where inventory_id<100
  • 89. More EXPLAIN Information http://www.pythian.com/news/wp-content/uploads/explain-diagram.pdf Pages 590 – 614 of the MySQL Administrator's Bible Sakila sample database: http://dev.mysql.com/doc/index-other.html
  • 90. Questions, comments, feedback? Sheeri K. Cabral – MySQL DBA www.oursql.com – MySQL podcast [email protected] @sheeri – send me questions, I may be able to help! Worked with MySQL since 2001 Full-time MySQL DBA as of 2005