如何正確的使用or展開來改寫SQL查詢

2024-07-21 02:42:16

字體：大中小

供稿：網(wǎng)友

問題:

下面的這條sql應(yīng)該怎么優(yōu)化?

select * from sys_user where user_code = 'zhangyong' or user_code in (select grp_code from sys_grp where sys_grp.user_code = 'zhangyong')Execution Plan----------------------------------------------------------0  SELECT STATEMENT Optimizer=RULE1  0   FILTER2  1     TABLE access (FULL) OF 'SYS_USER'3  1     INDEX (UNIQUE SCAN) OF 'PK_SYS_GRP' (UNIQUE)Statistics----------------------------------------------------------14  recursive calls4  db block gets30590 consistent gets0  physical reads0  redo size1723  bytes sent via SQL*Net to client425  bytes received via SQL*Net from client2  SQL*Net roundtrips to/from client0  sorts (memory)0  sorts (disk)3  rows PRocessed

里面的查詢返回的記錄數(shù)一般只有一兩條，但sys_user表的數(shù)據(jù)很多，怎么樣才能讓這條sql以sys_grp為驅(qū)動表？

表中記錄情況如下:

SQL> select count(*) from sys_grp;COUNT(*)----------25130SQL> select count(*) from sys_user;COUNT(*)----------15190

優(yōu)化:

降低邏輯讀是優(yōu)化SQL的基本原則之一

我們嘗試通過降低邏輯讀來加快SQL的執(zhí)行.

這里我們使用or展開來改寫SQL查詢:

select * from sys_user where user_code = 'zhangyong' union allselect * from sys_user where user_code <> 'zhangyong' and user_code in (select grp_code from sys_grp where sys_grp.user_code = 'zhangyong')Statistics----------------------------------------------------------          0  recursive calls          0  db block gets         130 consistent gets          0  physical reads          0  redo size       1723  bytes sent via SQL*Net to client        425  bytes received via SQL*Net from client          2  SQL*Net roundtrips to/from client          1  sorts (memory)          0  sorts (disk)          3  rows processedExecution Plan----------------------------------------------------------0      SELECT STATEMENT Optimizer=RULE1    0   UNION-ALL2    1     TABLE ACCESS (BY INDEX ROWID) OF 'SYS_USER'3    2       INDEX (UNIQUE SCAN) OF 'PK_SYS_USER' (UNIQUE)4    1     NESTED LOOPS5    4       VIEW OF 'VW_NSO_1'6    5         SORT (UNIQUE)7    6           TABLE ACCESS (BY INDEX ROWID) OF 'SYS_GRP'8    7             INDEX (RANGE SCAN) OF 'FK_SYS_USER_CODE' (NON-UNIQUE)9    4       TABLE ACCESS (BY INDEX ROWID) OF 'SYS_USER'10    9         INDEX (UNIQUE SCAN) OF 'PK_SYS_USER' (UNIQUE)

我們注意到,通過改寫,邏輯讀減少到130，從30590到130這是一個巨大的提高,減少邏輯讀最終會減少資源消耗,提高SQL的執(zhí)行效率.

這個改寫把Filter改為了Nest LOOP,索引得以充分利用.從而大大提高了性能.

我們同時注意到，這里引入了一個排序

排序來自于這一步:

----------------------------------6  5  SORT (UNIQUE)7  6    TABLE ACCESS (BY INDEX ROWID) OF 'SYS_GRP'8  7   INDEX (RANGE SCAN) OF 'FK_SYS_USER_CODE' (NON-UNIQUE)----------------------------------

注釋：在'SYS_GRP'表中，user_code 是非唯一鍵值，在in值判斷里，要做sort unique排序，去除重復(fù)值，這里的union all是不需要排序的。

上一篇：如何用dbms_rowid獲取rowid的詳細(xì)信息

下一篇：輕松掌握數(shù)據(jù)庫的具體分類及其相關(guān)概念