2、ndition>
SELECT
DISTINCT
ORDER BY
LIMIT
1、LIMIT 語句
分頁查詢是最常用的場景之一，但也通常也是最容易出問題的地方。比如對于下面簡單的語句，一般 DBA 想到的辦法是在 type, name, create_time 字段上加組合索引。這樣條件排序都能有效的利用到索引，性能迅速提升。
SELECT *
FROM operation
WHERE type = 'SQLStats'
AND
3、 name = 'SlowLog'
ORDER BY create_time
LIMIT 1000, 10;
好吧，可能90%以上的 DBA 解決該問題就到此為止。但當 LIMIT 子句變成 “LIMIT 1000000,10” 時，程序員仍然會抱怨：我只取10條記錄為什么還是慢？
要知道數(shù)據(jù)庫也并不知道第1000000條記錄從什么地方開始，即使有索引也需要從頭計算一次。出現(xiàn)這種性能問題，多數(shù)情形下是程序員偷懶了。
在前端數(shù)據(jù)瀏覽翻頁，或者大數(shù)據(jù)分批導(dǎo)出等場景下，是可以將上一頁的最大值當成參數(shù)作為查詢條件的。SQL 重新設(shè)計如下：
SELECT *
FROM
4、operation
WHERE type = 'SQLStats'
AND name = 'SlowLog'
AND create_time > '2017-03-16 14:00:00'
ORDER BY create_time limit 10;
在新設(shè)計下查詢時間基本固定，不會隨著數(shù)據(jù)量的增長而發(fā)生變化。
2、隱式轉(zhuǎn)換
SQL語句中查詢變量和字段定義類型不匹配是另一個常見的錯誤。比如下面的語句：
mysql> explain extended SELECT *
> FROM my_balance b
> WHERE
5、 b.bpn = 14000000123
> AND b.isverified IS NULL ;
mysql> show warnings;
| Warning | 1739 | Cannot use ref access on index 'bpn' due to type or collation conversion on field 'bpn'
其中字段 bpn 的定義為 varchar(20)，MySQL 的策略是將字符串轉(zhuǎn)換為數(shù)字之后再比較。函數(shù)作用于表字段，索引失效。
上述情況可能是應(yīng)用程序框架自動填入的參數(shù)，而不是程序員的原意。現(xiàn)在應(yīng)用框
6、架很多很繁雜，使用方便的同時也小心它可能給自己挖坑。
3、關(guān)聯(lián)更新、刪除
雖然 MySQL5.6 引入了物化特性，但需要特別注意它目前僅僅針對查詢語句的優(yōu)化。對于更新或刪除需要手工重寫成 JOIN。
比如下面 UPDATE 語句，MySQL 實際執(zhí)行的是循環(huán)/嵌套子查詢（DEPENDENT SUBQUERY)，其執(zhí)行時間可想而知。
UPDATE operation o
SET status = 'applying'
WHERE o.id IN (SELECT id
FROM (SELECT o.id,
7、 o.status
FROM operation o
WHERE o.group = 123
AND o.status NOT IN ( 'done' )
ORDER BY o.parent,
o.id
LIMIT 1) t);
8、
執(zhí)行計劃：
+----+--------------------+-------+-------+---------------+---------+---------+-------+------+-----------------------------------------------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra
9、 |
+----+--------------------+-------+-------+---------------+---------+---------+-------+------+-----------------------------------------------------+
| 1 | PRIMARY | o | index | | PRIMARY | 8 | | 24 | Using where; Using temporary
10、 |
| 2 | DEPENDENT SUBQUERY | | | | | | | | Impossible WHERE noticed after reading const tables |
| 3 | DERIVED | o | ref | idx_2,idx_5 | idx_5 | 8 | const | 1 | Using where; Using filesort
11、 |
+----+--------------------+-------+-------+---------------+---------+---------+-------+------+-----------------------------------------------------+
重寫為 JOIN 之后，子查詢的選擇模式從 DEPENDENT SUBQUERY 變成 DERIVED，執(zhí)行速度大大加快，從7秒降低到2毫秒。
UPDATE operation o
JOIN (SELECT o.id,
12、 o.status
FROM operation o
WHERE o.group = 123
AND o.status NOT IN ( 'done' )
ORDER BY o.parent,
o.id
LIMIT 1) t
ON o.id = t.
13、id
SET status = 'applying'
執(zhí)行計劃簡化為：
+----+-------------+-------+------+---------------+-------+---------+-------+------+-----------------------------------------------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra
14、 |
+----+-------------+-------+------+---------------+-------+---------+-------+------+-----------------------------------------------------+
| 1 | PRIMARY | | | | | | | | Impossible WHERE noticed after reading const tables
15、 |
| 2 | DERIVED | o | ref | idx_2,idx_5 | idx_5 | 8 | const | 1 | Using where; Using filesort |
+----+-------------+-------+------+---------------+-------+---------+-------+------+-----------------------------------------------------+
4、混合排序
MySQL
16、不能利用索引進行混合排序。但在某些場景，還是有機會使用特殊方法提升性能的。
SELECT *
FROM my_order o
INNER JOIN my_appraise a ON a.orderid = o.id
ORDER BY a.is_reply ASC,
a.appraise_time DESC
LIMIT 0, 20
執(zhí)行計劃顯示為全表掃描：
+----+-------------+-------+--------+-------------+---------+---------+---------------+-
17、--------+-+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra
+----+-------------+-------+--------+-------------+---------+---------+---------------+---------+-+
| 1 | SIMPLE | a | ALL | idx_orderid | NULL | NULL | NULL
18、| 1967647 | Using filesort |
| 1 | SIMPLE | o | eq_ref | PRIMARY | PRIMARY | 122 | a.orderid | 1 | NULL |
+----+-------------+-------+--------+---------+---------+---------+-----------------+---------+-+
由于 is_reply 只有0和1兩種狀態(tài)，我們按照下面的方法重寫后，執(zhí)行時間從1.58秒降低到2毫秒。
S
19、ELECT *
FROM ((SELECT *
FROM my_order o
INNER JOIN my_appraise a
ON a.orderid = o.id
AND is_reply = 0
ORDER BY appraise_time DESC
LIMIT 0, 20)
UNION ALL
(SELECT *
FR
20、OM my_order o
INNER JOIN my_appraise a
ON a.orderid = o.id
AND is_reply = 1
ORDER BY appraise_time DESC
LIMIT 0, 20)) t
ORDER BY is_reply ASC,
appraisetime DESC
LIMIT 20;
5、EXISTS語句
My
21、SQL 對待 EXISTS 子句時，仍然采用嵌套子查詢的執(zhí)行方式。如下面的 SQL 語句：
SELECT *
FROM my_neighbor n
LEFT JOIN my_neighbor_apply sra
ON n.id = sra.neighbor_id
AND sra.user_id = 'xxx'
WHERE n.topic_status < 4
AND EXISTS(SELECT 1
FROM message_info m
22、 WHERE n.id = m.neighbor_id
AND m.inuser = 'xxx')
AND n.topic_type <> 5
執(zhí)行計劃為：
+----+--------------------+-------+------+-----+------------------------------------------+---------+-------+---------+ -----+
| id | select_type | table | t
23、ype | possible_keys | key | key_len | ref | rows | Extra |
+----+--------------------+-------+------+ -----+------------------------------------------+---------+-------+---------+ -----+
| 1 | PRIMARY | n | ALL | | NULL | NULL | NULL | 1086041 | Using where
24、 |
| 1 | PRIMARY | sra | ref | | idx_user_id | 123 | const | 1 | Using where |
| 2 | DEPENDENT SUBQUERY | m | ref | | idx_message_info | 122 | const | 1 | Using index condition; Using where |
+----+--------------------+-------+
25、------+ -----+------------------------------------------+---------+-------+---------+ -----+
去掉 exists 更改為 join，能夠避免嵌套子查詢，將執(zhí)行時間從1.93秒降低為1毫秒。
SELECT *
FROM my_neighbor n
INNER JOIN message_info m
ON n.id = m.neighbor_id
AND m.inuser = 'xxx'
26、LEFT JOIN my_neighbor_apply sra
ON n.id = sra.neighbor_id
AND sra.user_id = 'xxx'
WHERE n.topic_status < 4
AND n.topic_type <> 5
新的執(zhí)行計劃：
+----+-------------+-------+--------+ -----+------------------------------------------+---------+ -----+------+ --
27、---+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+-------------+-------+--------+ -----+------------------------------------------+---------+ -----+------+ -----+
| 1 | SIMPLE | m | ref | | idx_message_in
28、fo | 122 | const | 1 | Using index condition |
| 1 | SIMPLE | n | eq_ref | | PRIMARY | 122 | ighbor_id | 1 | Using where |
| 1 | SIMPLE | sra | ref | | idx_user_id | 123 | const | 1 | Using where |
+----+-------------+-------+---
29、-----+ -----+------------------------------------------+---------+ -----+------+ -----+
6、條件下推
外部查詢條件不能夠下推到復(fù)雜的視圖或子查詢的情況有：
1、聚合子查詢；2、含有 LIMIT 的子查詢；3、UNION 或 UNION ALL 子查詢；4、輸出字段中的子查詢；
如下面的語句，從執(zhí)行計劃可以看出其條件作用于聚合子查詢之后：
SELECT *
FROM (SELECT target,
Count(*)
FROM ope
30、ration
GROUP BY target) t
WHERE target = 'rm-xxxx'
+----+-------------+------------+-------+---------------+-------------+---------+-------+------+-------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+--
31、-----------+------------+-------+---------------+-------------+---------+-------+------+-------------+
| 1 | PRIMARY | | ref | | | 514 | const | 2 | Using where |
| 2 | DERIVED | operation | index | idx_4 | idx_4 | 519 | NU
32、LL | 20 | Using index |
+----+-------------+------------+-------+---------------+-------------+---------+-------+------+-------------+
確定從語義上查詢條件可以直接下推后，重寫如下：
SELECT target,
Count(*)
FROM operation
WHERE target = 'rm-xxxx'
GROUP BY target
執(zhí)行計劃變?yōu)椋?
+----+-------------+--
33、---------+------+---------------+-------+---------+-------+------+--------------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+-------------+-----------+------+---------------+-------+---------+-------+------+--------------------+
| 1 | SIMP
34、LE | operation | ref | idx_4 | idx_4 | 514 | const | 1 | Using where; Using index |
+----+-------------+-----------+------+---------------+-------+---------+-------+------+--------------------+
關(guān)于 MySQL 外部條件不能下推的詳細解釋說明請參考以前文章：MySQL · 性能優(yōu)化 · 條件下推到物化表 http://mysql.taobao.org/monthly/2016/07/08
35、7、提前縮小范圍
先上初始 SQL 語句：
SELECT *
FROM my_order o
LEFT JOIN my_userinfo u
ON o.uid = u.uid
LEFT JOIN my_productinfo p
ON o.pid = p.pid
WHERE ( o.display = 0 )
AND ( o.ostaus = 1 )
ORDER BY o.selltime DESC
LIMIT 0, 15
該SQL語句原意是：先做一系列的
36、左連接，然后排序取前15條記錄。從執(zhí)行計劃也可以看出，最后一步估算排序記錄數(shù)為90萬，時間消耗為12秒。
+----+-------------+-------+--------+---------------+---------+---------+-----------------+--------+----------------------------------------------------+
| id | select_type | table | type | possible_keys | key | key_len | ref |
37、 rows | Extra |
+----+-------------+-------+--------+---------------+---------+---------+-----------------+--------+----------------------------------------------------+
| 1 | SIMPLE | o | ALL | NULL | NULL | NULL | NULL
38、 | 909119 | Using where; Using temporary; Using filesort |
| 1 | SIMPLE | u | eq_ref | PRIMARY | PRIMARY | 4 | o.uid | 1 | NULL |
| 1 | SIMPLE | p | ALL | PRIMARY | NULL | NULL | NULL
39、 | 6 | Using where; Using join buffer (Block Nested Loop) |
+----+-------------+-------+--------+---------------+---------+---------+-----------------+--------+----------------------------------------------------+
由于最后 WHERE 條件以及排序均針對最左主表，因此可以先對 my_order 排序提前縮小數(shù)據(jù)量再做左連接。SQL 重寫后如下，執(zhí)
40、行時間縮小為1毫秒左右。
SELECT *
FROM (
SELECT *
FROM my_order o
WHERE ( o.display = 0 )
AND ( o.ostaus = 1 )
ORDER BY o.selltime DESC
LIMIT 0, 15
) o
LEFT JOIN my_userinfo u
ON o.uid = u.uid
LEFT JOIN my_productinfo p
ON o.pid = p.pid
ORDER BY o
41、.selltime DESC
limit 0, 15
再檢查執(zhí)行計劃：子查詢物化后（select_type=DERIVED)參與 JOIN。雖然估算行掃描仍然為90萬，但是利用了索引以及 LIMIT 子句后，實際執(zhí)行時間變得很小。
+----+-------------+------------+--------+---------------+---------+---------+-------+--------+----------------------------------------------------+
| id | select_type | table
42、 | type | possible_keys | key | key_len | ref | rows | Extra |
+----+-------------+------------+--------+---------------+---------+---------+-------+--------+----------------------------------------------------+
| 1 | PRIMARY |
43、ed2> | ALL | NULL | NULL | NULL | NULL | 15 | Using temporary; Using filesort |
| 1 | PRIMARY | u | eq_ref | PRIMARY | PRIMARY | 4 | o.uid | 1 | NULL |
| 1 | PRIMARY | p
44、 | ALL | PRIMARY | NULL | NULL | NULL | 6 | Using where; Using join buffer (Block Nested Loop) |
| 2 | DERIVED | o | index | NULL | idx_1 | 5 | NULL | 909112 | Using where |
+----+-------------+-------
45、-----+--------+---------------+---------+---------+-------+--------+----------------------------------------------------+
8、中間結(jié)果集下推
再來看下面這個已經(jīng)初步優(yōu)化過的例子(左連接中的主表優(yōu)先作用查詢條件)：
SELECT a.*,
c.allocated
FROM (
SELECT resourceid
FROM my_distribute d
46、
WHERE isdelete = 0
AND cusmanagercode = '1234567'
ORDER BY salecode limit 20) a
LEFT JOIN
(
SELECT resourcesid， sum(ifnull(allocation, 0) * 12345) allocated
FROM my_resources
47、 GROUP BY resourcesid) c
ON a.resourceid = c.resourcesid
那么該語句還存在其它問題嗎？不難看出子查詢 c 是全表聚合查詢，在表數(shù)量特別大的情況下會導(dǎo)致整個語句的性能下降。
其實對于子查詢 c，左連接最后結(jié)果集只關(guān)心能和主表 resourceid 能匹配的數(shù)據(jù)。因此我們可以重寫語句如下，執(zhí)行時間從原來的2秒下降到2毫秒。
SELECT a.*,
c.allocated
FROM (
SELECT resou
48、rceid
FROM my_distribute d
WHERE isdelete = 0
AND cusmanagercode = '1234567'
ORDER BY salecode limit 20) a
LEFT JOIN
(
SELECT resourcesid， sum(ifnull(allocation, 0) * 12345)
49、allocated
FROM my_resources r,
(
SELECT resourceid
FROM my_distribute d
WHERE isdelete = 0
50、 AND cusmanagercode = '1234567'
ORDER BY salecode limit 20) a
WHERE r.resourcesid = a.resourcesid
GROUP BY resourcesid) c
ON a.resourceid = c.resourcesid
但是子查詢 a 在我們的SQL語句中出現(xiàn)了多次。這種寫法不僅存在額外的開銷，還使得整個語句
51、顯的繁雜。使用 WITH 語句再次重寫：
WITH a AS
(
SELECT resourceid
FROM my_distribute d
WHERE isdelete = 0
AND cusmanagercode = '1234567'
ORDER BY salecode limit 20)
SELECT a.*,
c.allocated
FROM a
LEFT JOIN
(
52、 SELECT resourcesid， sum(ifnull(allocation, 0) * 12345) allocated
FROM my_resources r,
a
WHERE r.resourcesid = a.resourcesid
GROUP BY resourcesid) c
ON a.resourceid = c.resourcesid
總結(jié)
數(shù)據(jù)庫編譯器產(chǎn)生執(zhí)行計劃，決定著SQL的實際執(zhí)行方式。但是編譯器只是盡力服務(wù)，所有數(shù)據(jù)庫的編譯器都不是盡善盡美的。
上述提到的多數(shù)場景，在其它數(shù)據(jù)庫中也存在性能問題。了解數(shù)據(jù)庫編譯器的特性，才能避規(guī)其短處，寫出高性能的SQL語句。
程序員在設(shè)計數(shù)據(jù)模型以及編寫SQL語句時，要把算法的思想或意識帶進來。
編寫復(fù)雜SQL語句要養(yǎng)成使用 WITH 語句的習慣。簡潔且思路清晰的SQL語句也能減小數(shù)據(jù)庫的負擔 。
SQL 書寫規(guī)范與各關(guān)鍵字詳解
