乱爱性全过程免费视频/午夜毛片视频/中文字幕av三区/一级视频网址

當(dāng)前位置： 維易PHP培訓(xùn)學(xué)院 > MYSQL教程 > 內(nèi)容正文

Mysql學(xué)習(xí)MySQL索引背后的之使用策略及優(yōu)化(高性能索引策略)

作者：VEPHP   時(shí)間 2017-10-03

《Mysql學(xué)習(xí)MySQL索引背后的之使用策略及優(yōu)化(高性能索引策略)》要點(diǎn)：
本文介紹了Mysql學(xué)習(xí)MySQL索引背后的之使用策略及優(yōu)化(高性能索引策略)，希望對(duì)您有用。如果有疑問，可以聯(lián)系我們。

本章的內(nèi)容完全基于上文的理論基礎(chǔ),實(shí)際上一旦理解了索引背后的機(jī)制,那么選擇高性能的策略就變成了純粹的推理,并且可以理解這些策略背后的邏輯.

　　示例數(shù)據(jù)庫MYSQL數(shù)據(jù)庫

　　為了討論索引策略,需要一個(gè)數(shù)據(jù)量不算小的數(shù)據(jù)庫作為示例.本文選用MySQL官方文檔中提供的示例數(shù)據(jù)庫之一：employees.這個(gè)數(shù)據(jù)庫關(guān)系復(fù)雜度適中,且數(shù)據(jù)量較大.下圖是這個(gè)數(shù)據(jù)庫的E-R關(guān)系圖(引用自MySQL官方手冊(cè))：MYSQL數(shù)據(jù)庫

　　MYSQL數(shù)據(jù)庫

\

?MYSQL數(shù)據(jù)庫

　　圖12MYSQL數(shù)據(jù)庫

　　MySQL官方文檔中關(guān)于此數(shù)據(jù)庫的頁面為http://dev.mysql.com/doc/employee/en/employee.html.里面詳細(xì)介紹了此數(shù)據(jù)庫,并提供了下載地址和導(dǎo)入方法,如果有興趣導(dǎo)入此數(shù)據(jù)庫到自己的MySQL可以參考文中內(nèi)容.MYSQL數(shù)據(jù)庫

　　最左前綴原理與相關(guān)優(yōu)化MYSQL數(shù)據(jù)庫

　　高效使用索引的首要條件是知道什么樣的查詢會(huì)使用到索引,這個(gè)問題和B+Tree中的“最左前綴原理”有關(guān),下面通過例子說明最左前綴原理.MYSQL數(shù)據(jù)庫

　　這里先說一下聯(lián)合索引的概念.在上文中,我們都是假設(shè)索引只引用了單個(gè)的列,實(shí)際上,MySQL中的索引可以以一定順序引用多個(gè)列,這種索引叫做聯(lián)合索引,一般的,一個(gè)聯(lián)合索引是一個(gè)有序元組,其中各個(gè)元素均為數(shù)據(jù)表的一列,實(shí)際上要嚴(yán)格定義索引需要用到關(guān)系代數(shù),但是這里我不想討論太多關(guān)系代數(shù)的話題,因?yàn)槟菢訒?huì)顯得很枯燥,所以這里就不再做嚴(yán)格定義.另外,單列索引可以看成聯(lián)合索引元素?cái)?shù)為1的特例.MYSQL數(shù)據(jù)庫

　　以employees.titles表為例,下面先查看其上都有哪些索引：MYSQL數(shù)據(jù)庫

SHOW INDEX FROM employees.titles;
+--------+------------+----------+--------------+-------------+-----------+-------------+------+------------+
| Table? | Non_unique | Key_name | Seq_in_index | Column_name | Collation | Cardinality | Null | Index_type |
+--------+------------+----------+--------------+-------------+-----------+-------------+------+------------+
| titles |????????? 0 | PRIMARY? |??????????? 1 | emp_no????? | A???????? |??????? NULL |????? | BTREE????? |
| titles |????????? 0 | PRIMARY? |??????????? 2 | title?????? | A???????? |??????? NULL |????? | BTREE????? |
| titles |????????? 0 | PRIMARY? |??????????? 3 | from_date?? | A???????? |????? 443308 |????? | BTREE????? |
| titles |????????? 1 | emp_no?? |??????????? 1 | emp_no????? | A???????? |????? 443308 |????? | BTREE????? |
+--------+------------+----------+--------------+-------------+-----------+-------------+------+------------+ MYSQL數(shù)據(jù)庫

?MYSQL數(shù)據(jù)庫

　　從結(jié)果中可以到titles表的主索引為,還有一個(gè)輔助索引.為了避免多個(gè)索引使事情變復(fù)雜(MySQL的SQL優(yōu)化器在多索引時(shí)行為比較復(fù)雜),這里我們將輔助索引drop掉：,>MYSQL數(shù)據(jù)庫

　　ALTER TABLE employees.titles DROP INDEX emp_no;MYSQL數(shù)據(jù)庫

　　這樣就可以專心分析索引PRIMARY的行為了.MYSQL數(shù)據(jù)庫

?MYSQL數(shù)據(jù)庫

　　情況一：全列匹配.MYSQL數(shù)據(jù)庫

EXPLAIN SELECT * FROM employees.titles WHERE emp_no='10001' AND title='Senior Engineer' AND from_date='1986-06-26';
+----+-------------+--------+-------+---------------+---------+---------+-------------------+------+-------+
| id | select_type | table? | type? | possible_keys | key???? | key_len | ref?????????????? | rows | Extra |
+----+-------------+--------+-------+---------------+---------+---------+-------------------+------+-------+
|? 1 | SIMPLE????? | titles | const | PRIMARY?????? | PRIMARY | 59????? | const,const,const |??? 1 |?????? |
+----+-------------+--------+-------+---------------+---------+---------+-------------------+------+-------+ MYSQL數(shù)據(jù)庫

?MYSQL數(shù)據(jù)庫

　　很明顯,當(dāng)按照索引中所有列進(jìn)行精確匹配(這里精確匹配指“=”或“IN”匹配)時(shí),索引可以被用到.這里有一點(diǎn)需要注意,理論上索引對(duì)順序是敏感的,但 是由于MySQL的查詢優(yōu)化器會(huì)自動(dòng)調(diào)整where子句的條件順序以使用適合的索引,例如我們將where中的條件順序顛倒：MYSQL數(shù)據(jù)庫

EXPLAIN SELECT * FROM employees.titles WHERE from_date='1986-06-26' AND emp_no='10001' AND title='Senior Engineer';
+----+-------------+--------+-------+---------------+---------+---------+-------------------+------+-------+
| id | select_type | table? | type? | possible_keys | key???? | key_len | ref?????????????? | rows | Extra |
+----+-------------+--------+-------+---------------+---------+---------+-------------------+------+-------+
|? 1 | SIMPLE????? | titles | const | PRIMARY?????? | PRIMARY | 59????? | const,const,const |??? 1 |?????? |
+----+-------------+--------+-------+---------------+---------+---------+-------------------+------+-------+ MYSQL數(shù)據(jù)庫

?MYSQL數(shù)據(jù)庫

　　效果是一樣的.MYSQL數(shù)據(jù)庫

　　情況二：最左前綴匹配.MYSQL數(shù)據(jù)庫

EXPLAIN SELECT * FROM employees.titles WHERE emp_no='10001';
+----+-------------+--------+------+---------------+---------+---------+-------+------+-------+
| id | select_type | table? | type | possible_keys | key???? | key_len | ref?? | rows | Extra |
+----+-------------+--------+------+---------------+---------+---------+-------+------+-------+
|? 1 | SIMPLE????? | titles | ref? | PRIMARY?????? | PRIMARY | 4?????? | const |??? 1 |?????? |
+----+-------------+--------+------+---------------+---------+---------+-------+------+-------+ MYSQL數(shù)據(jù)庫

?MYSQL數(shù)據(jù)庫

　　當(dāng)查詢條件精確匹配索引的左邊連續(xù)一個(gè)或幾個(gè)列時(shí),如或,所以可以被用到,但是只能用到一部分,即條件所組成的最左前綴.上面的查詢從分析結(jié)果看用到了PRIMARY索引,但是 key_len為4,說明只用到了索引的第一列前綴.,>MYSQL數(shù)據(jù)庫

　　情況三：查詢條件用到了索引中列的精確匹配,但是中間某個(gè)條件未提供.MYSQL數(shù)據(jù)庫

EXPLAIN SELECT * FROM employees.titles WHERE emp_no='10001' AND from_date='1986-06-26';
+----+-------------+--------+------+---------------+---------+---------+-------+------+-------------+
| id | select_type | table? | type | possible_keys | key???? | key_len | ref?? | rows | Extra?????? |
+----+-------------+--------+------+---------------+---------+---------+-------+------+-------------+
|? 1 | SIMPLE????? | titles | ref? | PRIMARY?????? | PRIMARY | 4?????? | const |??? 1 | Using where |
+----+-------------+--------+------+---------------+---------+---------+-------+------+-------------+ MYSQL數(shù)據(jù)庫

?MYSQL數(shù)據(jù)庫

　　此時(shí)索引使用情況和情況二相同,因?yàn)閠itle未提供,所以查詢只用到了索引的第一列,而后面的from_date雖然也在索引中,但是由于 title不存在而無法和左前綴連接,因此需要對(duì)結(jié)果進(jìn)行掃描過濾from_date(這里由于emp_no唯一,所以不存在掃描).如果想讓 from_date也使用索引而不是where過濾,可以增加一個(gè)輔助索引,此時(shí)上面的查詢會(huì)使用這個(gè)索引.除此之外,還可以使用一種稱之為“隔離列”的優(yōu)化方法,將emp_no與from_date 之間的“坑”填上.,>MYSQL數(shù)據(jù)庫

　　首先我們看下title一共有幾種不同的值：MYSQL數(shù)據(jù)庫

SELECT DISTINCT(title) FROM employees.titles;
+--------------------+
| title????????????? |
+--------------------+
| Senior Engineer??? |
| Staff????????????? |
| Engineer?????????? |
| Senior Staff?????? |
| Assistant Engineer |
| Technique Leader?? |
| Manager??????????? |
+--------------------+ MYSQL數(shù)據(jù)庫

?MYSQL數(shù)據(jù)庫

　　只有7種.在這種成為“坑”的列值比較少的情況下,可以考慮用“IN”來填補(bǔ)這個(gè)“坑”從而形成最左前綴：MYSQL數(shù)據(jù)庫

EXPLAIN SELECT * FROM employees.titles
WHERE emp_no='10001'
AND title IN ('Senior Engineer', 'Staff', 'Engineer', 'Senior Staff', 'Assistant Engineer', 'Technique Leader', 'Manager')
AND from_date='1986-06-26';
+----+-------------+--------+-------+---------------+---------+---------+------+------+-------------+
| id | select_type | table? | type? | possible_keys | key???? | key_len | ref? | rows | Extra?????? |
+----+-------------+--------+-------+---------------+---------+---------+------+------+-------------+
|? 1 | SIMPLE????? | titles | range | PRIMARY?????? | PRIMARY | 59????? | NULL |??? 7 | Using where |
+----+-------------+--------+-------+---------------+---------+---------+------+------+-------------+ MYSQL數(shù)據(jù)庫

?MYSQL數(shù)據(jù)庫

　　這次key_len為59,說明索引被用全了,但是從type和rows看出IN實(shí)際上執(zhí)行了一個(gè)range查詢,這里檢查了7個(gè)key.看下兩種查詢的性能比較：MYSQL數(shù)據(jù)庫

SHOW PROFILES;
+----------+------------+-------------------------------------------------------------------------------+
| Query_ID | Duration?? | Query???????????????????????????????????????????????????????????????????????? |
+----------+------------+-------------------------------------------------------------------------------+
|?????? 10 | 0.00058000 | SELECT * FROM employees.titles WHERE emp_no='10001' AND from_date='1986-06-26'|
|?????? 11 | 0.00052500 | SELECT * FROM employees.titles WHERE emp_no='10001' AND title IN ...????????? |
+----------+------------+-------------------------------------------------------------------------------+ MYSQL數(shù)據(jù)庫

?MYSQL數(shù)據(jù)庫

　　“填坑”后性能提升了一點(diǎn).如果經(jīng)過emp_no篩選后余下很多數(shù)據(jù),則后者性能優(yōu)勢(shì)會(huì)更加明顯.當(dāng)然,如果title的值很多,用填坑就不合適了,必須建立輔助索引.MYSQL數(shù)據(jù)庫

　　情況四：查詢條件沒有指定索引第一列.MYSQL數(shù)據(jù)庫

EXPLAIN SELECT * FROM employees.titles WHERE from_date='1986-06-26';??????????????????
+----+-------------+--------+------+---------------+------+---------+------+--------+-------------+
| id | select_type | table? | type | possible_keys | key? | key_len | ref? | rows?? | Extra?????? |
+----+-------------+--------+------+---------------+------+---------+------+--------+-------------+
|? 1 | SIMPLE????? | titles | ALL? | NULL????????? | NULL | NULL??? | NULL | 443308 | Using where |
+----+-------------+--------+------+---------------+------+---------+------+--------+-------------+ MYSQL數(shù)據(jù)庫

?MYSQL數(shù)據(jù)庫

　　由于不是最左前綴,索引這樣的查詢顯然用不到索引.MYSQL數(shù)據(jù)庫

　　情況五：匹配某列的前綴字符串.MYSQL數(shù)據(jù)庫

EXPLAIN SELECT * FROM employees.titles WHERE emp_no='10001' AND title LIKE 'Senior%';
view sourceprint?
+----+-------------+--------+-------+---------------+---------+---------+------+------+-------------+
| id | select_type | table? | type? | possible_keys | key???? | key_len | ref? | rows | Extra?????? |
+----+-------------+--------+-------+---------------+---------+---------+------+------+-------------+
|? 1 | SIMPLE????? | titles | range | PRIMARY?????? | PRIMARY | 56????? | NULL |??? 1 | Using where |
+----+-------------+--------+-------+---------------+---------+---------+------+------+-------------+ MYSQL數(shù)據(jù)庫

?MYSQL數(shù)據(jù)庫

　　此時(shí)可以用到索引,但是如果通配符不是只出現(xiàn)在末尾,則無法使用索引.MYSQL數(shù)據(jù)庫

　　情況六：范圍查詢.MYSQL數(shù)據(jù)庫

EXPLAIN SELECT * FROM employees.titles WHERE emp_no<'10010' and title='Senior Engineer';
+----+-------------+--------+-------+---------------+---------+---------+------+------+-------------+
| id | select_type | table? | type? | possible_keys | key???? | key_len | ref? | rows | Extra?????? |
+----+-------------+--------+-------+---------------+---------+---------+------+------+-------------+
|? 1 | SIMPLE????? | titles | range | PRIMARY?????? | PRIMARY | 4?????? | NULL |?? 16 | Using where |
+----+-------------+--------+-------+---------------+---------+---------+------+------+-------------+ MYSQL數(shù)據(jù)庫

?MYSQL數(shù)據(jù)庫

　　范圍列可以用到索引(必須是最左前綴),但是范圍列后面的列無法用到索引.同時(shí),索引最多用于一個(gè)范圍列,因此如果查詢條件中有兩個(gè)范圍列則無法全用到索引.MYSQL數(shù)據(jù)庫

EXPLAIN SELECT * FROM employees.titles
WHERE emp_no<'10010'
AND title='Senior Engineer'
AND from_date BETWEEN '1986-01-01' AND '1986-12-31';
+----+-------------+--------+-------+---------------+---------+---------+------+------+-------------+
| id | select_type | table? | type? | possible_keys | key???? | key_len | ref? | rows | Extra?????? |
+----+-------------+--------+-------+---------------+---------+---------+------+------+-------------+
|? 1 | SIMPLE????? | titles | range | PRIMARY?????? | PRIMARY | 4?????? | NULL |?? 16 | Using where |
+----+-------------+--------+-------+---------------+---------+---------+------+------+-------------+ MYSQL數(shù)據(jù)庫

?MYSQL數(shù)據(jù)庫

　　可以看到索引對(duì)第二個(gè)范圍索引無能為力.這里特別要說明MySQL一個(gè)有意思的地方,那就是僅用explain可能無法區(qū)分范圍索引和多值匹配,因?yàn)樵趖ype中這兩者都顯示為range.同時(shí),用了“between”并不意味著就是范圍查詢,例如下面的查詢：MYSQL數(shù)據(jù)庫

EXPLAIN SELECT * FROM employees.titles
WHERE emp_no BETWEEN '10001' AND '10010'
AND title='Senior Engineer'
AND from_date BETWEEN '1986-01-01' AND '1986-12-31';
+----+-------------+--------+-------+---------------+---------+---------+------+------+-------------+
| id | select_type | table? | type? | possible_keys | key???? | key_len | ref? | rows | Extra?????? |
+----+-------------+--------+-------+---------------+---------+---------+------+------+-------------+
|? 1 | SIMPLE????? | titles | range | PRIMARY?????? | PRIMARY | 59????? | NULL |?? 16 | Using where |
+----+-------------+--------+-------+---------------+---------+---------+------+------+-------------+ MYSQL數(shù)據(jù)庫

?MYSQL數(shù)據(jù)庫

　　看起來是用了兩個(gè)范圍查詢,但作用于emp_no上的“BETWEEN”實(shí)際上相當(dāng)于“IN”,也就是說emp_no實(shí)際是多值精確匹配.可以看到這個(gè)查詢用到了索引全部三個(gè)列.因此在MySQL中要謹(jǐn)慎地區(qū)分多值匹配和范圍匹配,否則會(huì)對(duì)MySQL的行為產(chǎn)生困惑.MYSQL數(shù)據(jù)庫

　　情況七：查詢條件中含有函數(shù)或表達(dá)式.MYSQL數(shù)據(jù)庫

　　很不幸,如果查詢條件中含有函數(shù)或表達(dá)式,則MySQL不會(huì)為這列使用索引(雖然某些在數(shù)學(xué)意義上可以使用).例如：MYSQL數(shù)據(jù)庫

EXPLAIN SELECT * FROM employees.titles WHERE emp_no='10001' AND left(title, 6)='Senior';
+----+-------------+--------+------+---------------+---------+---------+-------+------+-------------+
| id | select_type | table? | type | possible_keys | key???? | key_len | ref?? | rows | Extra?????? |
+----+-------------+--------+------+---------------+---------+---------+-------+------+-------------+
|? 1 | SIMPLE????? | titles | ref? | PRIMARY?????? | PRIMARY | 4?????? | const |??? 1 | Using where |
+----+-------------+--------+------+---------------+---------+---------+-------+------+-------------+ MYSQL數(shù)據(jù)庫

?MYSQL數(shù)據(jù)庫

　　雖然這個(gè)查詢和情況五中功能相同,但是由于使用了函數(shù)left,則無法為title列應(yīng)用索引,而情況五中用LIKE則可以.再如：MYSQL數(shù)據(jù)庫

EXPLAIN SELECT * FROM employees.titles WHERE emp_no - 1='10000';???????????????????????
+----+-------------+--------+------+---------------+------+---------+------+--------+-------------+
| id | select_type | table? | type | possible_keys | key? | key_len | ref? | rows?? | Extra?????? |
+----+-------------+--------+------+---------------+------+---------+------+--------+-------------+
|? 1 | SIMPLE????? | titles | ALL? | NULL????????? | NULL | NULL??? | NULL | 443308 | Using where |
+----+-------------+--------+------+---------------+------+---------+------+--------+-------------+ MYSQL數(shù)據(jù)庫

?MYSQL數(shù)據(jù)庫

　　顯然這個(gè)查詢等價(jià)于查詢emp_no為10001的函數(shù),但是由于查詢條件是一個(gè)表達(dá)式,MySQL無法為其使用索引.看來MySQL還沒有智能到自動(dòng)優(yōu)化常量表達(dá)式的程度,因此在寫查詢語句時(shí)盡量避免表達(dá)式出現(xiàn)在查詢中,而是先手工私下代數(shù)運(yùn)算,轉(zhuǎn)換為無表達(dá)式的查詢語句.MYSQL數(shù)據(jù)庫

?MYSQL數(shù)據(jù)庫

　　索引選擇性與前綴索引MYSQL數(shù)據(jù)庫

　　既然索引可以加快查詢速度,那么是不是只要是查詢語句需要,就建上索引?答案是否定的.因?yàn)樗饕m然加快了查詢速度,但索引也是有代價(jià)的：索引文件本身要消耗存儲(chǔ)空間,同時(shí)索引會(huì)加重插入、刪除和修改記錄時(shí)的負(fù)擔(dān),另外,MySQL在運(yùn)行時(shí)也要消耗資源維護(hù)索引,因此索引并不是越多越好.一般兩種情況下不建議建索引.MYSQL數(shù)據(jù)庫

　　第一種情況是表記錄比較少,例如一兩千條甚至只有幾百條記錄的表,沒必要建索引,讓查詢做全表掃描就好了.至于多少條記錄才算多,這個(gè)個(gè)人有個(gè)人的看法,我個(gè)人的經(jīng)驗(yàn)是以2000作為分界線,記錄數(shù)不超過 2000可以考慮不建索引,超過2000條可以酌情考慮索引.MYSQL數(shù)據(jù)庫

　　另一種不建議建索引的情況是索引的選擇性較低.所謂索引的選擇性(Selectivity),是指不重復(fù)的索引值(也叫基數(shù),Cardinality)與表記錄數(shù)(#T)的比值：MYSQL數(shù)據(jù)庫

　　Index Selectivity = Cardinality / #TMYSQL數(shù)據(jù)庫

　　顯然選擇性的取值范圍為(0, 1],選擇性越高的索引價(jià)值越大,這是由B+Tree的性質(zhì)決定的.例如,上文用到的employees.titles表,如果title字段經(jīng)常被單獨(dú)查詢,是否需要建索引,我們看一下它的選擇性：MYSQL數(shù)據(jù)庫

SELECT count(DISTINCT(title))/count(*) AS Selectivity FROM employees.titles;
+-------------+
| Selectivity |
+-------------+
|????? 0.0000 |
+-------------+MYSQL數(shù)據(jù)庫

?MYSQL數(shù)據(jù)庫

　　title的選擇性不足0.0001(精確值為0.00001579),所以實(shí)在沒有什么必要為其單獨(dú)建索引.MYSQL數(shù)據(jù)庫

　　有一種與索引選擇性有關(guān)的索引優(yōu)化策略叫做前綴索引,就是用列的前綴代替整個(gè)列作為索引key,當(dāng)前綴長度合適時(shí),可以做到既使得前綴索引的選擇性 接近全列索引,同時(shí)因?yàn)樗饕齥ey變短而減少了索引文件的大小和維護(hù)開銷.下面以employees.employees表為例介紹前綴索引的選擇和使 用.MYSQL數(shù)據(jù)庫

　　從圖12可以看到employees表只有一個(gè)索引,那么如果我們想按名字搜索一個(gè)人,就只能全表掃描了：MYSQL數(shù)據(jù)庫

EXPLAIN SELECT * FROM employees.employees WHERE first_name='Eric' AND last_name='Anido';????????????????
+----+-------------+-----------+------+---------------+------+---------+------+--------+-------------+
| id | select_type | table???? | type | possible_keys | key? | key_len | ref? | rows?? | Extra?????? |
+----+-------------+-----------+------+---------------+------+---------+------+--------+-------------+
|? 1 | SIMPLE????? | employees | ALL? | NULL????????? | NULL | NULL??? | NULL | 300024 | Using where |
+----+-------------+-----------+------+---------------+------+---------+------+--------+-------------+ MYSQL數(shù)據(jù)庫

?MYSQL數(shù)據(jù)庫

　　如果頻繁按名字搜索員工,這樣顯然效率很低,因此我們可以考慮建索引.有兩種選擇,建或,看下兩個(gè)索引的選擇性：,>MYSQL數(shù)據(jù)庫

SELECT count(DISTINCT(first_name))/count(*) AS Selectivity FROM employees.employees;
+-------------+
| Selectivity |
+-------------+
|????? 0.0042 |
+-------------+

SELECT count(DISTINCT(concat(first_name, last_name)))/count(*) AS Selectivity FROM employees.employees;
+-------------+
| Selectivity |
+-------------+
|????? 0.9313 |
+-------------+ MYSQL數(shù)據(jù)庫

??? <first_name>顯然選擇性太低,<first_name, last_name>選擇性很好,但是first_name和last_name加起來長度為30,有沒有兼顧長度和選擇性的辦法?可以考慮用 first_name和last_name的前幾個(gè)字符建立索引,例如<first_name, left(last_name, 3)>,看看其選擇性：MYSQL數(shù)據(jù)庫

SELECT count(DISTINCT(concat(first_name, left(last_name, 3))))/count(*) AS Selectivity FROM employees.employees;
+-------------+
| Selectivity |
+-------------+
|????? 0.7879 |
+-------------+ MYSQL數(shù)據(jù)庫

?MYSQL數(shù)據(jù)庫

　　選擇性還不錯(cuò),但離0.9313還是有點(diǎn)距離,那么把last_name前綴加到4：MYSQL數(shù)據(jù)庫

SELECT count(DISTINCT(concat(first_name, left(last_name, 4))))/count(*) AS Selectivity FROM employees.employees;
+-------------+
| Selectivity |
+-------------+
|????? 0.9007 |
+-------------+ MYSQL數(shù)據(jù)庫

?MYSQL數(shù)據(jù)庫

　　這時(shí)選擇性已經(jīng)很理想了,而這個(gè)索引的長度只有18,比短了接近一半,我們把這個(gè)前綴索引<first_name, last_name>,>建上：MYSQL數(shù)據(jù)庫

ALTER TABLE employees.employees
ADD INDEX `first_name_last_name4` (first_name, last_name(4)); MYSQL數(shù)據(jù)庫

?MYSQL數(shù)據(jù)庫

　　此時(shí)再執(zhí)行一遍按名字查詢,比較分析一下與建索引前的結(jié)果：MYSQL數(shù)據(jù)庫

SHOW PROFILES;
+----------+------------+---------------------------------------------------------------------------------+
| Query_ID | Duration?? | Query?????????????????????????????????????????????????????????????????????????? |
+----------+------------+---------------------------------------------------------------------------------+
|?????? 87 | 0.11941700 | SELECT * FROM employees.employees WHERE first_name='Eric' AND last_name='Anido' |
|?????? 90 | 0.00092400 | SELECT * FROM employees.employees WHERE first_name='Eric' AND last_name='Anido' |
+----------+------------+---------------------------------------------------------------------------------+ MYSQL數(shù)據(jù)庫

?MYSQL數(shù)據(jù)庫

　　性能的提升是顯著的,查詢速度提高了120多倍.MYSQL數(shù)據(jù)庫

　　前綴索引兼顧索引大小和查詢速度,但是其缺點(diǎn)是不能用于ORDER BY和GROUP BY操作,也不能用于Covering index(即當(dāng)索引本身包含查詢所需全部數(shù)據(jù)時(shí),不再訪問數(shù)據(jù)文件本身).MYSQL數(shù)據(jù)庫

　　InnoDB的主鍵選擇與插入優(yōu)化MYSQL數(shù)據(jù)庫

　　在使用InnoDB存儲(chǔ)引擎時(shí),如果沒有特別的需要,請(qǐng)永遠(yuǎn)使用一個(gè)與業(yè)務(wù)無關(guān)的自增字段作為主鍵.MYSQL數(shù)據(jù)庫

　　經(jīng)常看到有帖子或博客討論主鍵選擇問題,有人建議使用業(yè)務(wù)無關(guān)的自增主鍵,有人覺得沒有必要,完全可以使用如學(xué)號(hào)或身份證號(hào)這種唯一字段作為主鍵.不論支持哪種論點(diǎn),大多數(shù)論據(jù)都是業(yè)務(wù)層面的.如果從數(shù)據(jù)庫索引優(yōu)化角度看,使用InnoDB引擎而不使用自增主鍵絕對(duì)是一個(gè)糟糕的主意.MYSQL數(shù)據(jù)庫

　　上文討論過InnoDB的索引實(shí)現(xiàn),InnoDB使用聚集索引,數(shù)據(jù)記錄本身被存于主索引(一顆B+Tree)的葉子節(jié)點(diǎn)上.這就要求同一個(gè)葉子節(jié)點(diǎn)內(nèi)(大小為一個(gè)內(nèi)存頁或磁盤頁)的各條數(shù)據(jù)記錄按主鍵順序存放,因此每當(dāng)有一條新的記錄插入時(shí),MySQL會(huì)根據(jù)其主鍵將其插入適當(dāng)?shù)墓?jié)點(diǎn)和位置,如果頁面達(dá)到裝載因子(InnoDB默認(rèn)為15/16),則開辟一個(gè)新的頁(節(jié)點(diǎn)).MYSQL數(shù)據(jù)庫

　　如果表使用自增主鍵,那么每次插入新的記錄,記錄就會(huì)順序添加到當(dāng)前索引節(jié)點(diǎn)的后續(xù)位置,當(dāng)一頁寫滿,就會(huì)自動(dòng)開辟一個(gè)新的頁.如下圖所示：MYSQL數(shù)據(jù)庫

　　MYSQL數(shù)據(jù)庫

\

?MYSQL數(shù)據(jù)庫

　　圖13MYSQL數(shù)據(jù)庫

　　這樣就會(huì)形成一個(gè)緊湊的索引結(jié)構(gòu),近似順序填滿.由于每次插入時(shí)也不需要移動(dòng)已有數(shù)據(jù),因此效率很高,也不會(huì)增加很多開銷在維護(hù)索引上.MYSQL數(shù)據(jù)庫

　　如果使用非自增主鍵(如果身份證號(hào)或?qū)W號(hào)等),由于每次插入主鍵的值近似于隨機(jī),因此每次新紀(jì)錄都要被插到現(xiàn)有索引頁得中間某個(gè)位置：MYSQL數(shù)據(jù)庫

　　MYSQL數(shù)據(jù)庫

\

?MYSQL數(shù)據(jù)庫

　　圖14MYSQL數(shù)據(jù)庫

　　此時(shí)MySQL不得不為了將新記錄插到合適位置而移動(dòng)數(shù)據(jù),甚至目標(biāo)頁面可能已經(jīng)被回寫到磁盤上而從緩存中清掉,此時(shí)又要從磁盤上讀回來,這增加了很多開銷,同時(shí)頻繁的移動(dòng)、分頁操作造成了大量的碎片,得到了不夠緊湊的索引結(jié)構(gòu),后續(xù)不得不通過OPTIMIZE TABLE來重建表并優(yōu)化填充頁面.MYSQL數(shù)據(jù)庫

　　因此,只要可以,請(qǐng)盡量在InnoDB上采用自增字段做主鍵.MYSQL數(shù)據(jù)庫

轉(zhuǎn)載請(qǐng)注明本頁網(wǎng)址：
http://www.fzlkiss.com/jiaocheng/6417.html


    標(biāo)簽：