`
dengyll
  • 浏览: 90358 次
社区版块
存档分类
最新评论

count(*)与count(col)对比

oracle 
阅读更多

Masql的MyISAM引擎 count(*)与count(col)对比

不带where的条件下:

使用的是Server version: 5.1.34 MySQL Community Server (GPL)

使用count(*)执行结果

mysql> select count(*) from my_cms_25;                 
+----------+
| count(*) |
+----------+
| 1022711 |
+----------+
1 row in set (0.00 sec)

mysql> show profiles;
+----------+------------+--------------------------------+
| Query_ID | Duration   | Query                          |
+----------+------------+--------------------------------+
|        1 | 0.00027000 | select count(*) from my_cms_25 |
+----------+------------+--------------------------------+

使用count(id)执行结果:

mysql> select count(id) from my_cms_25;
+-----------+
| count(id) |
+-----------+
|   1022711 |
+-----------+
1 row in set (0.00 sec)

mysql> show profiles;
+----------+------------+---------------------------------+
| Query_ID | Duration   | Query                           |
+----------+------------+---------------------------------+
|        1 | 0.00027800 | select count(id) from my_cms_25 |
+----------+------------+---------------------------------+
1 row in set (0.00 sec)

为了保证cache,每次执行一条语句,就把mysql重启动.

mysql> select count(x_rank) from my_cms_25;
+---------------+
| count(x_rank) |
+---------------+
|       1022711 |
+---------------+
1 row in set (0.00 sec)

mysql> show profiles;
+----------+------------+-------------------------------------+
| Query_ID | Duration   | Query                               |
+----------+------------+-------------------------------------+
|        1 | 0.00029200 | select count(x_rank) from my_cms_25 |
+----------+------------+-------------------------------------+
1 row in set (0.00 sec)

count(*)比其他两种都快, 不管是count(*), count(id)还是count(x_rank) 执行时间差的不是太多.

注: id是primary key, x_rank是非index.

 

 

带where从句:

mysql> select count(*) from my_cms_25 where id > 100000;
+----------+
| count(*) |
+----------+
|   922711 |
+----------+
1 row in set (0.55 sec)

mysql> show profiles;
+----------+------------+--------------------------------------------------+
| Query_ID | Duration   | Query                                            |
+----------+------------+--------------------------------------------------+
|        1 | 0.54154100 | select count(*) from my_cms_25 where id > 100000 |
+----------+------------+--------------------------------------------------+
1 row in set (0.00 sec)

mysql> select count(id) from my_cms_25 where id > 100000;
+-----------+
| count(id) |
+-----------+
|    922711 |
+-----------+
1 row in set (0.52 sec)

mysql> show profiles;
+----------+------------+---------------------------------------------------+
| Query_ID | Duration   | Query                                             |
+----------+------------+---------------------------------------------------+
|        1 | 0.51955100 | select count(id) from my_cms_25 where id > 100000 |
+----------+------------+---------------------------------------------------+
1 row in set (0.00 sec)

 

mysql> select count(*) from my_cms_25 where id > 100;
+----------+
| count(*) |
+----------+
| 1022611 |
+----------+
1 row in set (0.60 sec)

mysql> show profiles;
+----------+------------+-----------------------------------------------+
| Query_ID | Duration   | Query                                         |
+----------+------------+-----------------------------------------------+
|        1 | 0.59619000 | select count(*) from my_cms_25 where id > 100 |
+----------+------------+-----------------------------------------------+
1 row in set (0.00 sec)

mysql> select count(id) from my_cms_25 where id > 100;
+-----------+
| count(id) |
+-----------+
|   1022611 |
+-----------+
1 row in set (0.57 sec)

mysql> show profiles;
+----------+------------+------------------------------------------------+
| Query_ID | Duration   | Query                                          |
+----------+------------+------------------------------------------------+
|        1 | 0.57422700 | select count(id) from my_cms_25 where id > 100 |
+----------+------------+------------------------------------------------+
1 row in set (0.00 sec)

mysql> select count(x_rank) from my_cms_25 where id > 100;
+---------------+
| count(x_rank) |
+---------------+
|       1022611 |
+---------------+
1 row in set (1.31 sec)

mysql> show profiles;
+----------+------------+----------------------------------------------------+
| Query_ID | Duration   | Query                                              |
+----------+------------+----------------------------------------------------+
|        1 | 1.30890800 | select count(x_rank) from my_cms_25 where id > 100 |
+----------+------------+----------------------------------------------------+
1 row in set (0.00 sec)

带where条件的, 测试的结果count(id) > count(*) > count(x_rank)

就是说,count(*)不一定比count(pk)快.

 

 

对于innodb engine的, 叶兄做了一个测试,自己也做了一个测试,测试结果不相同:

http://imysql.cn/2008_06_24_speedup_innodb_count

http://bugs.mysql.com/bug.php?id=19271

 

使用的版本:Server version: 5.1.30-log Source distribution

mysql> select count(*) from relation;
+----------+
| count(*) |
+----------+
| 3010500 |
+----------+
1 row in set (10.60 sec)

mysql> show profiles;
+----------+-------------+-------------------------------+
| Query_ID | Duration    | Query                         |
+----------+-------------+-------------------------------+
|        1 | 10.59423300 | select count(*) from relation |
+----------+-------------+-------------------------------+
1 row in set (0.01 sec)

 

 

mysql> select count(id) from relation;
+-----------+
| count(id) |
+-----------+
|   3010329 |
+-----------+
1 row in set (3.87 sec)

mysql> show profiles;
+----------+------------+--------------------------------+
| Query_ID | Duration   | Query                          |
+----------+------------+--------------------------------+
|        1 | 3.87004700 | select count(id) from relation |
+----------+------------+--------------------------------+
1 row in set (0.00 sec)

 

 

mysql> select count(*) from relation where id>100;
+----------+
| count(*) |
+----------+
| 3010158 |
+----------+
1 row in set (1.03 sec)

mysql> show profiles;                
+----------+------------+--------------------------------------------+
| Query_ID | Duration   | Query                                      |
+----------+------------+--------------------------------------------+
|        1 | 3.87004700 | select count(id) from relation             |
|        2 | 1.03079800 | select count(*) from relation where id>100 |
+----------+------------+--------------------------------------------+

 

 

mysql> show profiles;
+----------+------------+--------------------------------------------------+
| Query_ID | Duration   | Query                                            |
+----------+------------+--------------------------------------------------+
|        1 | 3.87004700 | select count(id) from relation                   |
|        2 | 1.03079800 | select count(*) from relation where id>100       |
|        3 | 1.01789300 | select count(id) from relation where id>100      |
|        4 | 1.32853100 | select count(*) from relation where infoid > 100 |
+----------+------------+--------------------------------------------------+

 

 

 

 

测试结果不全相同. 不管带不带where条件时count(pk)是比count(*)快的．

count(secondary key) 是没有count(pk)快的。难道bug fix了在5.1.30?

 

感觉应该是pk的存储结构和secondary key不同.这个和官方说的符合,primary key的scan是最快的原因吧.

 

从mysql query optimizer角度上说, count(*) 需要mysql分析更多的column,这个是有一定的overhead的。

但直接count(pk)就没有分析的这个开销.:)

 

下面是更详细,可以看出慢在那一步:
在myisam engine上:

 

 

 

mysql> select count(*) from my_cms_25;
+----------+
| count(*) |
+----------+
| 1022711 |
+----------+
1 row in set (0.00 sec)

mysql> show profile;
+--------------------------------+----------+
| Status                         | Duration |
+--------------------------------+----------+
| starting                       | 0.000033 |
| checking query cache for query | 0.000088 |
| Opening tables                 | 0.000014 |
| System lock                    | 0.000005 |
| Table lock                     | 0.000026 |
| init                           | 0.000033 |
| optimizing                     | 0.000013 |
| executing                      | 0.000015 |
| end                            | 0.000004 |
| query end                      | 0.000003 |
| freeing items                  | 0.000024 |
| storing result in query cache | 0.000012 |
| logging slow query             | 0.000003 |
| cleaning up                    | 0.000004 |
+--------------------------------+----------+
14 rows in set (0.00 sec)

mysql> show profiles;
+----------+------------+--------------------------------+
| Query_ID | Duration   | Query                          |
+----------+------------+--------------------------------+
|        1 | 0.00027700 | select count(*) from my_cms_25 |
+----------+------------+--------------------------------+
1 row in set (0.00 sec)

 

 

mysql> select count(id) from my_cms_25;
+-----------+
| count(id) |
+-----------+
|   1022711 |
+-----------+
1 row in set (0.00 sec)

mysql> show profile;
+--------------------------------+----------+
| Status                         | Duration |
+--------------------------------+----------+
| starting                       | 0.000056 |
| checking query cache for query | 0.000077 |
| Opening tables                 | 0.000014 |
| System lock                    | 0.000005 |
| Table lock                     | 0.000025 |
| init                           | 0.000047 |
| optimizing                     | 0.000014 |
| executing                      | 0.000015 |
| end                            | 0.000004 |
| query end                      | 0.000003 |
| freeing items                  | 0.000027 |
| storing result in query cache | 0.000011 |
| logging slow query             | 0.000004 |
| cleaning up                    | 0.000003 |
+--------------------------------+----------+
14 rows in set (0.00 sec)

mysql> show profiles;
+----------+------------+---------------------------------+
| Query_ID | Duration   | Query                           |
+----------+------------+---------------------------------+
|        1 | 0.00030500 | select count(id) from my_cms_25 |
+----------+------------+---------------------------------+
1 row in set (0.00 sec)

 

 

innodb engine:

 

mysql> select count(*) from sbtest;
+----------+
| count(*) |
+----------+
| 1000000 |
+----------+
1 row in set (1.98 sec)

mysql> show profile;
+--------------------------------+----------+
| Status                         | Duration |
+--------------------------------+----------+
| starting                       | 0.000041 |
| checking query cache for query | 0.000082 |
| Opening tables                 | 0.000015 |
| System lock                    | 0.000004 |
| Table lock                     | 0.000042 |
| init                           | 0.000033 |
| optimizing                     | 0.000012 |
| statistics                     | 0.000014 |
| preparing                      | 0.000010 |
| executing                      | 0.000009 |
| Sending data                   | 1.987284 |
| end                            | 0.000019 |
| query end                      | 0.000010 |
| freeing items                  | 0.000078 |
| storing result in query cache | 0.000016 |
| logging slow query             | 0.000005 |
| cleaning up                    | 0.000008 |
+--------------------------------+----------+
17 rows in set (0.00 sec)

mysql> show profiles;
+----------+------------+-----------------------------+
| Query_ID | Duration   | Query                       |
+----------+------------+-----------------------------+
|        1 | 1.98768200 | select count(*) from sbtest |
+----------+------------+-----------------------------+
1 row in set (0.00 sec)

 

 

 

mysql> select count(id) from sbtest;
+-----------+
| count(id) |
+-----------+
|   1000000 |
+-----------+
1 row in set (1.99 sec)

mysql> show profile;
+--------------------------------+----------+
| Status                         | Duration |
+--------------------------------+----------+
| starting                       | 0.000034 |
| checking query cache for query | 0.000079 |
| Opening tables                 | 0.000015 |
| System lock                    | 0.000005 |
| Table lock                     | 0.000042 |
| init                           | 0.000037 |
| optimizing                     | 0.000013 |
| statistics                     | 0.000013 |
| preparing                      | 0.000010 |
| executing                      | 0.000021 |
| Sending data                   | 1.982898 |
| end                            | 0.000021 |
| query end                      | 0.000009 |
| freeing items                  | 0.000089 |
| storing result in query cache | 0.000017 |
| logging slow query             | 0.000006 |
| cleaning up                    | 0.000010 |
+--------------------------------+----------+
17 rows in set (0.00 sec)

mysql> show profiles;
+----------+------------+------------------------------+
| Query_ID | Duration   | Query                        |
+----------+------------+------------------------------+
|        1 | 1.98331900 | select count(id) from sbtest |
+----------+------------+------------------------------+
1 row in set (0.00 sec)

 

分享到：
| 数据库系统依然很神秘
评论
发表评论

您还没有登录,请您登录后再发表评论

相关推荐

    MySQL中count(*)、count(1)和count(col)的区别汇总

    count()函数是用来统计表中记录的一个函数，返回匹配条件的行数，下面这篇文章主要给大家总结介绍了关于MySQL中count(*)、count(1)和count(col)的区别，文中通过示例代码介绍的非常详细，需要的朋友可以参考下。

    帐套不能启用出现数据库问题 重复键

    设表名为X,字段为C1,C2,C3,.... ...SELECT COUNT(*) AS RowNumber,COL1,COL2 FROM TABLE1 T1 WHERE (SELECT COUNT(*) FROM TABLE1 T2 WHERE T1.COL1 = T2.COL1 AND T1.COL2 = T2.COL2)>1 GROUP BY COL1,COL2

    MySQL优化总结-查询总条数

    1、COUNT(*)和COUNT(COL) COUNT(*)通常是对主键进行索引扫描，而COUNT(COL)就不一定了，另外前者是统计表中的所有符合的纪录总数，而后者是计算表中所有符合的COL的纪录数。还有有区别的。 优化总结，对于MyISAM表...

    mysql技巧之select count的区别分析

    (3)select count(col_name)方式 分别使用 select count(group_id) select count(user_id) select count(col_null)  通过上述测试结果可以看到，select count(*)和select count(1)都使用了group_id这个最短

    SQL2000_SQL2005_SQL2008语句大全word完整版

    十分好的资源 建议大家一定掌握 例如 -合并处理 SELECT col1, col2=CAST(MIN(col2)as varchar) ... WHEN COUNT(*)>=2 THEN ','+CAST(MAX(col2)as varchar) ELSE '' END FROM tb a GROUP BY col1 DROP TABLE tb

    linux c语言操作数据库(连接sqlite数据库)

    int select_callback(void *data,int col_count,char **col_values,char **col_name){ //每条记录回调一次该函数，有多少条就回调多少次 int i; for(i=0;i<col_count;i++) { printf(“%s=%s\n”,col_name[i]...

    #include

    min=mt[j].row*mt[j].col*mt[j+step].col +rows[j][0].cost+rows[j+1][step-1].cost;//__page_break__ mid=j; for(i=1;i;i++) { temp=rows[j][i].cost+rows[j+i+1][step-i-1].cost +mt[j].row*mt[j+i].col*mt[j+step...

    机器人的运动范围.pdf

    count = 1 + movingCountCore(threshold, rows, cols, row + 1, col, visited) + movingCountCore(threshold, rows, cols, row, col + 1, visited) + movingCountCore(threshold, rows, cols, row - 1, col, ...

    jsp/java/数据库实现分页

    select count(*) as id from User1 2、每页显示几条记录 由自己规定 3、得出一共能分为几页 (allCol+ipage-1)/ipage 4、当前页为第几页 默认值为1，随着变量的变化而变化 5、每页的开始位置和结束位置 指的是结果...

    扫雷.cpp（部分代码）

    else if(k==m_col*(m_row-1)) { scan_mine(k+1); scan_mine(k-m_col); scan_mine(k-m_col+1); } else if(k==m_col*m_row-1) { scan_mine(k-1); scan_mine(k-m_col); scan_mine(k-m_col-1); } else if(k...

    T-SQL高级查询

    可以完成对分组的数据进行增加排序，partition by可以与以上三个函数联合使用。 利用partition by按照班级名称分组，学生id排序 select s.id, s.name, cid, c.name, row_number() over(partition by c.name order ...

    扫雷中布雷算法源码.txt

    int count = 0, row1, col1; srand(time_t(0));//time do{ row = rand() % 10; col = rand() % 10; printf("%d,%d\n", row, col); if (arr[row][col] == 0){//排除重复的可能 arr[row][col] = -1; ...

    SQL操作全集(非常适合初学者)

    总数：select count * as totalcount from table1 求和：select sum(field1) as sumvalue from table1 平均：select avg(field1) as avgvalue from table1 最大：select max(field1) as maxvalue from table1 ...

    KEY_4x4扫描键盘FPGAVerilog逻辑源码Quartus工程文件.rar

    count <= count + 1; end end always @(posedge clk_500khz or negedge sys_rst_n) if(!sys_rst_n) begin key_col; state; end else begin case (state) 0: begin key_col[3:0]; ...

    ZENCARTzencart首页图片错乱的修改

    $col_width = floor(100/$num_products_count); } else { $col_width = floor(100/SHOW_PRODUCT_INFO_COLUMNS_NEW_PRODUCTS); } 改为 $col_width = floor(100/$num_products_count-1); } else { $col_width = floor...

    PHP PDOStatement::columnCount讲解

    PDOStatement::columnCount PDOStatement::columnCount — 返回结果集中的列数。(PHP 5 >= 5.1.0, PECL pdo >= 0.2.0) 说明 语法 int PDOStatement::columnCount ( void ) 使用PDOStatement::columnCount()返回由 ...

    调色板提取秘密信息试验结果

    调色板程序及其结果1.嵌入秘密信息 clc; clear; %读入载体图像 [x,map] = imread('lenaindex.bmp','bmp'); wx=x;...[row col]=size(wx);...[msg,count]=fread(msgfid);...count=24;...for i=1:count ...[row col]=ran

    sql函数和事务

    RETURNS @returnTable table(id int, col_Value varchar(50)) AS BEGIN declare @thisSplitStr varchar(50) declare @thisSepIndex int declare @lastSepIndex int declare @i int set @lastSepIndex = 0 set @i ...

    Aspose.cells最新7.4版本破解 可使用C#操作Excel

    cell sheet Cells[rowIndex col + columnIndex]; cell PutValue dt Columns[col] ColumnName ; } } private void AddBody DataTable dt int rowIndex int columnIndex { for int r 0; r < dt Rows Count...

Magicbox
Global site tag (gtag.js) - Google Analytics