count(*)，count(1)，count(字段)的對比

原創 jzzw 2019-10-25 22:11

Oracle：

參考 https://cloud.tencent.com/developer/article/1388976的方法，使用10053進行分析。
使用的版本爲Oracle 11.2.0.4

SQL> alter session set events='10053 trace name context forever,level 2';

會話已更改。

SQL> create table test1012(id1 number primary key,id2 number,id3 number);

表已創建。

SQL> create unique index idx_1012_3 on test1012(id3);

索引已創建。

SQL> declare
  2  begin
  3    for i in 1 .. 10000 loop
  4      insert into test1012 values (i, i + 1, i + 2);
  5    end loop;
  6    commit;
  7  end;
  8  /

PL/SQL 過程已成功完成。

SQL> select count(*) from test.test1012;

  COUNT(*)
----------
     10000

SQL> select count(1) from test.test1012;

  COUNT(1)
----------
     10000

SQL> select count(id1) from test.test1012;

COUNT(ID1)
----------
     10000

SQL> select count(id2) from test.test1012;

COUNT(ID2)
----------
      9999

SQL> select count(id3) from test.test1012;

COUNT(ID3)
----------
     10000

然後去看看trace文件(只截取部分) 

*******************************************
Peeked values of the binds in SQL statement
*******************************************

Final query after transformations:******* UNPARSED QUERY IS *******
SELECT COUNT(*) "COUNT(*)" FROM "TEST"."TEST1012" "TEST1012"
kkoqbc: optimizing query block SEL$1 (#0)
        
        :
    call(in-use=1216, alloc=16344), compile(in-use=56264, alloc=58632), execution(in-use=2464, alloc=4032)

kkoqbc-subheap (create addr=0x0000000000E7FA10)
============
Plan Table
============
---------------------------------------------+-----------------------------------+
| Id  | Operation              | Name        | Rows  | Bytes | Cost  | Time      |
---------------------------------------------+-----------------------------------+
| 0   | SELECT STATEMENT       |             |       |       |     7 |           |
| 1   |  SORT AGGREGATE        |             |     1 |       |       |           |
| 2   |   INDEX FAST FULL SCAN | SYS_C0011149|   10K |       |     7 |  00:00:01 |
---------------------------------------------+-----------------------------------+


CNT:   Considering count(col) to count(*) on query block SEL$1 (#0)
*************************
Count(col) to Count(*) (CNT)
*************************
CNT:     COUNT() to COUNT(*) done.
query block SEL$1 (#0) unchanged
Considering Query Transformations on query block SEL$1 (#0)
*******************************************
Peeked values of the binds in SQL statement
*******************************************

Final query after transformations:******* UNPARSED QUERY IS *******
SELECT COUNT(*) "COUNT(1)" FROM "TEST"."TEST1012" "TEST1012"
kkoqbc: optimizing query block SEL$1 (#0)
        
        :
    call(in-use=1240, alloc=16344), compile(in-use=56384, alloc=58632), execution(in-use=2464, alloc=4032)

kkoqbc-subheap (create addr=0x0000000000E7FA10)
============
Plan Table
============
---------------------------------------------+-----------------------------------+
| Id  | Operation              | Name        | Rows  | Bytes | Cost  | Time      |
---------------------------------------------+-----------------------------------+
| 0   | SELECT STATEMENT       |             |       |       |     7 |           |
| 1   |  SORT AGGREGATE        |             |     1 |       |       |           |
| 2   |   INDEX FAST FULL SCAN | SYS_C0011149|   10K |       |     7 |  00:00:01 |
---------------------------------------------+-----------------------------------+
** Using dynamic sampling card. : 10000
** Dynamic sampling updated table card.
  Table: TEST1012  Alias: TEST1012
    Card: Original: 10000.000000  Rounded: 10000  Computed: 10000.00  Non Adjusted: 10000.00
  Access Path: TableScan
    Cost:  9.05  Resp: 9.05  Degree: 0
      Cost_io: 9.00  Cost_cpu: 1699400
      Resp_io: 9.00  Resp_cpu: 1699400
  Access Path: index (index (FFS))
    Index: SYS_C0011149
    resc_io: 7.00  resc_cpu: 442429
    ix_sel: 0.000000  ix_sel_with_filters: 1.000000 
  Access Path: index (FFS)
    Cost:  7.01  Resp: 7.01  Degree: 1
      Cost_io: 7.00  Cost_cpu: 442429
      Resp_io: 7.00  Resp_cpu: 442429
  Access Path: index (FullScan)
    Index: SYS_C0011149
    resc_io: 21.00  resc_cpu: 649550
    ix_sel: 1.000000  ix_sel_with_filters: 1.000000 
    Cost: 21.02  Resp: 21.02  Degree: 1
******** Begin index join costing ********
  ****** trying bitmap/domain indexes ******
  Access Path: index (FullScan)
    Index: SYS_C0011149
    resc_io: 21.00  resc_cpu: 649550
    ix_sel: 1.000000  ix_sel_with_filters: 1.000000 
    Cost: 21.02  Resp: 21.02  Degree: 0
  Access Path: index (FullScan)
    Index: SYS_C0011149
    resc_io: 21.00  resc_cpu: 649550
    ix_sel: 1.000000  ix_sel_with_filters: 1.000000 
    Cost: 21.02  Resp: 21.02  Degree: 0
  Bitmap nodes:
    Used SYS_C0011149
      Cost = 26.273311, sel = 1.000000
  ****** finished trying bitmap/domain indexes ******
******** End index join costing ********
  Best:: AccessPath: IndexFFS
  Index: SYS_C0011149
         Cost: 7.01  Degree: 1  Resp: 7.01  Card: 10000.00  Bytes: 0

***************************************


CNT:   Considering count(col) to count(*) on query block SEL$1 (#0)
*************************
Count(col) to Count(*) (CNT)
*************************
CNT:     Converting COUNT(ID1) to COUNT(*).
CNT:     COUNT() to COUNT(*) done.
query block SEL$1 (#0) unchanged
*******************************************
Peeked values of the binds in SQL statement
*******************************************

Final query after transformations:******* UNPARSED QUERY IS *******
SELECT COUNT(*) "COUNT(ID1)" FROM "TEST"."TEST1012" "TEST1012"
kkoqbc: optimizing query block SEL$1 (#0)
        
        :
    call(in-use=1240, alloc=16344), compile(in-use=56752, alloc=58632), execution(in-use=2464, alloc=4032)

kkoqbc-subheap (create addr=0x0000000000E7FA10)
============
Plan Table
============
---------------------------------------------+-----------------------------------+
| Id  | Operation              | Name        | Rows  | Bytes | Cost  | Time      |
---------------------------------------------+-----------------------------------+
| 0   | SELECT STATEMENT       |             |       |       |     7 |           |
| 1   |  SORT AGGREGATE        |             |     1 |       |       |           |
| 2   |   INDEX FAST FULL SCAN | SYS_C0011149|   10K |       |     7 |  00:00:01 |
---------------------------------------------+-----------------------------------+



CNT:   Considering count(col) to count(*) on query block SEL$1 (#0)
*************************
Count(col) to Count(*) (CNT)
*************************
CNT:     COUNT() to COUNT(*) not done.
query block SEL$1 (#0) unchanged
*******************************************
Peeked values of the binds in SQL statement
*******************************************

Final query after transformations:******* UNPARSED QUERY IS *******
SELECT COUNT("TEST1012"."ID2") "COUNT(ID2)" FROM "TEST"."TEST1012" "TEST1012"
kkoqbc: optimizing query block SEL$1 (#0)
        
        :
    call(in-use=1240, alloc=16344), compile(in-use=56752, alloc=58632), execution(in-use=2464, alloc=4032)

kkoqbc-subheap (create addr=0x0000000000E7FA10)
============
Plan Table
============
--------------------------------------+-----------------------------------+
| Id  | Operation           | Name    | Rows  | Bytes | Cost  | Time      |
--------------------------------------+-----------------------------------+
| 0   | SELECT STATEMENT    |         |       |       |     9 |           |
| 1   |  SORT AGGREGATE     |         |     1 |    13 |       |           |
| 2   |   TABLE ACCESS FULL | TEST1012|   10K |  127K |     9 |  00:00:01 |
--------------------------------------+-----------------------------------+


CNT:   Considering count(col) to count(*) on query block SEL$1 (#0)
*************************
Count(col) to Count(*) (CNT)
*************************
CNT:     COUNT() to COUNT(*) not done.
query block SEL$1 (#0) unchanged
*******************************************
Peeked values of the binds in SQL statement
*******************************************

Final query after transformations:******* UNPARSED QUERY IS *******
SELECT COUNT("TEST1012"."ID3") "COUNT(ID3)" FROM "TEST"."TEST1012" "TEST1012"
kkoqbc: optimizing query block SEL$1 (#0)
        
        :
    call(in-use=1240, alloc=16344), compile(in-use=56752, alloc=58632), execution(in-use=2464, alloc=4032)

kkoqbc-subheap (create addr=0x0000000000E7FA10)
============
Plan Table
============
-------------------------------------------+-----------------------------------+
| Id  | Operation              | Name      | Rows  | Bytes | Cost  | Time      |
-------------------------------------------+-----------------------------------+
| 0   | SELECT STATEMENT       |           |       |       |     7 |           |
| 1   |  SORT AGGREGATE        |           |     1 |    13 |       |           |
| 2   |   INDEX FAST FULL SCAN | IDX_1012_3|   10K |  127K |     7 |  00:00:01 |
-------------------------------------------+-----------------------------------+

       我們可以發現，每一步都有一個 CNT:   Considering count(col) to count(*)，是不是Oracle在做這種count()聚合查詢時，都在嘗試能不能轉換成count(*)，如果可以的話就直接將sql進行轉換，例如：

SELECT COUNT(*) "COUNT(1)" FROM "TEST"."TEST1012" "TEST1012"

       看似count(1),實則count(*)。理所當然的，他們的執行計劃也是一樣的，都是對主鍵的index fast full scan

       所以如果只是爲了統計行數，我們可以認爲count(*)=count(1)=count(主鍵)
       count(字段)的話，更多的是業務上的需求(去掉null值的統計)，因爲有個篩選的過程，所以相比來說肯定是要慢點。

Mysql：

       我們這裏就只討論InnoDB了。我的Mysql版本是5.7.19。
       和Oracle的區別是，InnoDB的表是IOT，所以對於Mysql來說，二級索引上的信息也足夠我們進行count()的，在count(1)和count(*)時，Mysql會選取它認爲最快的遍歷方式，比如我這實驗：

mysql> show create table emp3;
+-------+-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| Table | Create Table                                                                                                                                                                                                                                    |
+-------+-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
| emp3  | CREATE TABLE `emp3` (
  `id` int(10) NOT NULL,
  `name` varchar(25) DEFAULT NULL,
  `deptid` int(11) DEFAULT NULL,
  `salary` float DEFAULT NULL,
  PRIMARY KEY (`id`),
  KEY `idx_emp3_salary` (`salary`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8 |
+-------+-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+
1 row in set (0.02 sec)

mysql> desc select count(id) from emp3;
+----+-------------+-------+------------+-------+---------------+-----------------+---------+------+--------+----------+-------------+
| id | select_type | table | partitions | type  | possible_keys | key             | key_len | ref  | rows   | filtered | Extra       |
+----+-------------+-------+------------+-------+---------------+-----------------+---------+------+--------+----------+-------------+
|  1 | SIMPLE      | emp3  | NULL       | index | NULL          | idx_emp3_salary | 5       | NULL | 997998 |   100.00 | Using index |
+----+-------------+-------+------------+-------+---------------+-----------------+---------+------+--------+----------+-------------+
1 row in set, 1 warning (0.01 sec)

mysql> desc select count(*) from emp3;
+----+-------------+-------+------------+-------+---------------+-----------------+---------+------+--------+----------+-------------+
| id | select_type | table | partitions | type  | possible_keys | key             | key_len | ref  | rows   | filtered | Extra       |
+----+-------------+-------+------------+-------+---------------+-----------------+---------+------+--------+----------+-------------+
|  1 | SIMPLE      | emp3  | NULL       | index | NULL          | idx_emp3_salary | 5       | NULL | 997998 |   100.00 | Using index |
+----+-------------+-------+------------+-------+---------------+-----------------+---------+------+--------+----------+-------------+
1 row in set, 1 warning (0.01 sec)

mysql> desc select count(1) from emp3;
+----+-------------+-------+------------+-------+---------------+-----------------+---------+------+--------+----------+-------------+
| id | select_type | table | partitions | type  | possible_keys | key             | key_len | ref  | rows   | filtered | Extra       |
+----+-------------+-------+------------+-------+---------------+-----------------+---------+------+--------+----------+-------------+
|  1 | SIMPLE      | emp3  | NULL       | index | NULL          | idx_emp3_salary | 5       | NULL | 997998 |   100.00 | Using index |
+----+-------------+-------+------------+-------+---------------+-----------------+---------+------+--------+----------+-------------+
1 row in set, 1 warning (0.00 sec)

mysql> desc select count(name) from emp3;
+----+-------------+-------+------------+------+---------------+------+---------+------+--------+----------+-------+
| id | select_type | table | partitions | type | possible_keys | key  | key_len | ref  | rows   | filtered | Extra |
+----+-------------+-------+------------+------+---------------+------+---------+------+--------+----------+-------+
|  1 | SIMPLE      | emp3  | NULL       | ALL  | NULL          | NULL | NULL    | NULL | 997998 |   100.00 | NULL  |
+----+-------------+-------+------------+------+---------------+------+---------+------+--------+----------+-------+
1 row in set, 1 warning (0.00 sec)

mysql> desc select count(salary) from emp3;
+----+-------------+-------+------------+-------+---------------+-----------------+---------+------+--------+----------+-------------+
| id | select_type | table | partitions | type  | possible_keys | key             | key_len | ref  | rows   | filtered | Extra       |
+----+-------------+-------+------------+-------+---------------+-----------------+---------+------+--------+----------+-------------+
|  1 | SIMPLE      | emp3  | NULL       | index | NULL          | idx_emp3_salary | 5       | NULL | 997998 |   100.00 | Using index |
+----+-------------+-------+------------+-------+---------------+-----------------+---------+------+--------+----------+-------------+

       可以看到，Mysql選擇了遍歷idx_emp3_salary這個二級索引。
       對於InnoDB來說，在遍歷之後，InnoDB會把結果取出來，返回給Server層，然後由Server層進行非空判斷並累加。
       但是 count(*) 是例外，並不會把全部字段取出來，而是專門做了優化，不取值。
       (參考：https://time.geekbang.org/column/article/72775)

       從官方文檔上我們也可以看到：
       Prior to MySQL 5.7.18, InnoDB processes SELECT COUNT(*) statements by scanning the clustered index. As of MySQL 5.7.18, InnoDB processes SELECT COUNT(*) statements by traversing the smallest available secondary index unless an index or optimizer hint directs the optimizer to use a different index. If a secondary index is not present, the clustered index is scanned.
       InnoDB handles SELECT COUNT(*) and SELECT COUNT(1) operations in the same way. There is no performance difference.

       所以對於InnoDB，我們可以認爲count(*)≈count(1)>count(主鍵)>count(字段)

 

       

 

 

發表評論
所有評論
還沒有人評論，想成為第一個評論的人麼? 請在上方評論欄輸入並且點擊發布.
相關文章

linux CentOS7 mysql 源碼安裝和rpm安裝

今天是安裝linux mysql第三天，終於安裝成功了。 第一種方法，源碼安裝。 下載 cmake-2.8.10.2.tar.gz 和 mysql-5.6.4-m7.tar.gz 上傳到服務器解壓完成 1.在cmake-2.8.10.2這

weixin_38081382 2020-07-08 12:31:47

linux mysql ERROR 1820 (HY000): Unknown error 1820

安裝好mysql登錄進去mysql中，使用查詢，創建命令都不行，還報錯。 網上查找後，發現雖然使用我上一遍文章修改了密碼，但是mysql中並沒有設置密碼，需要在mysql中再設置一下密碼  執行 SET PASSWORD = PASSWO

weixin_38081382 2020-07-08 12:31:47

Java之mySQL數據庫事務隔離級別

問題 同一個應用程序中的多個事務或不同應用程序中的多個事務在同一個數據集上併發執行時, 可能會出現許多意外的問題，這些問題可分爲如下三種類型： 髒讀(Drity Read)： 已知有兩個事務A和B, A讀取了已經被B更新但還沒

这个名字真难起 2020-07-08 12:03:47

Java之mySQL數據庫表

字段類型 整數類型 tinyInt：很小的整數， smallint：小的整數 mediumint：中等大小的整數 int：普通大小的整數 小數類型 float(m,d)：單精度浮點數，m表示數字長度，d表示小數位數，例如floa

这个名字真难起 2020-07-08 12:03:46

oracle存儲過程使用MD5加密

md5Pwd := Utl_Raw.Cast_To_Raw(sys.dbms_obfuscation_toolkit.md5(input_string => pwd)); md5Pwd:自己定義的變量，用來接收加密好了的字符串 pwd:

dreamZhf 2020-07-08 12:36:08

oracle查詢IP地域信息

declare url varchar2(2000); req utl_http.req; resp utl_http.resp; value VARCHAR2(1000); region VARCHAR2(50)

dreamZhf 2020-07-08 12:36:08

mysql預處理語句提取變量，實現動態執行SQL語句

爲了項目在長期使用過程中，數據庫仍能夠保持較好的性能，考慮將數據庫表分爲當前表和歷史表。當前錶針對不同客戶，分別存儲在不同的表中。因爲需要在mysql的存儲過程中實現動態的指向不同的表進行操作。 首先,mysql的預處理語句如果帶參數，每

pengdayong77 2020-07-08 12:37:23

面試準備季——MyBatis 面試專題（含答案）

寫在前面：2020年面試必備的Java後端進階面試題總結了一份複習指南在Github上，內容詳細，圖文並茂，有需要學習的朋友可以Star一下！ GitHub地址：https://github.com/abel-max/Java-S

毛发旺盛的程序员 2020-07-08 12:27:20

Flex+java開發環境配置

     最近在做個項目。前端是用flex來寫，後臺用java，數據庫就用mysql。定了以後開始配環境。網上這方面的資料很多都是不負責任的，照着來的話也會有不少的問題，把自己的配置方法寫下來，希望需要的人能用得上。  需要下載的軟件 M

duran1986 2020-07-08 12:24:31

MySQL：CentOS 7通過yum安裝MariaDB

如何在CentOS 7下通過yum安裝MariaDB呢？ 這裏附上大佬的安裝步驟： https://zhuanlan.zhihu.com/p/49046496 博主就是按照這個步驟來的，炒雞詳細~相信你也能看明白 |ू･ω･` )

ETalien_ 2020-07-08 12:19:01

MySQL：外鍵及外鍵的使用

外鍵及外鍵的使用 如果公共關鍵字在一個關係中是主關鍵字，那麼這個公共關鍵字被稱爲另一個關係的外鍵。 由此可見，外鍵表示了兩個關係之間的相關聯繫。以另一個關係的外鍵作主關鍵字的表被稱爲主表，具有此外鍵的表被稱爲主表的從表。外鍵又稱作

ETalien_ 2020-07-08 12:19:01

MySQL：建庫建表等簡單操作

庫的相關操作 – 查看已有庫 SHOW DATABASES; SHOW DATABASES; SHOW DATABASES\G – 新建庫的操作 CREATE DATABASE 庫名; CREATE DATABASE `ET`;

ETalien_ 2020-07-08 12:19:01

linux 下安裝 mysql5.6.30

以下安裝步驟是本人以前操作的過程步驟，不保證100%正確，如果在使用過程出現異常請查閱其他相關資料。 1、下載mysql5.6，地址：http://dev.mysql.com/downloads/mysql/ 我們這下載：mysql

华华鱼 2020-07-08 12:16:40

【數據庫-MySQL】mysql日誌詳解及性能優化

mysql日誌分爲四大類：慢查詢日誌，查詢日誌，錯誤日誌，二進制日誌 一.慢查詢日誌：大於慢查詢時間的語句，都會記錄在慢查詢日誌中 1.查看慢查詢是否開啓 show variables like "slow_query_log"; 如下：

斯台普斯的少年 2020-07-08 12:15:14

Mysql疑難雜症收集

  Mysql疑難雜症收集 By Eagoo 　　　　一、可疑問題： 　　　　　　　　mysql>show processlist; 　　　　

单线程的娃 2020-07-08 12:00:56