1:全表前100W条的所有数据
----------------------------------------------------
[SQL] select * from e_enterprise limit 1000000
影响的数据栏: 0
时间: 38.219ms
2:全表前100W条的id数据
----------------------------------------------------
[SQL] select id from e_enterprise limit 1000000
影响的数据栏: 0
时间: 1.187ms
3:全表前100W条的count(id)数据
----------------------------------------------------
[SQL] select count(id) from e_enterprise limit 1000000
影响的数据栏: 0
时间: 0.563ms
4:全表前100W条的count(*)数据
----------------------------------------------------
[SQL] select count(*) from e_enterprise limit 1000000
影响的数据栏: 0
时间: 0.531ms
数据实时的不一致,即使数据不变的查询的时间也不一样
在印象中count(id)总比count(*)要快的,但是我看了一下测试结果竟然count(*)更快,id是一个主键,怎么会这样呢?
下面是测试数据,数据说明了一切
----------------------------------------------------
[SQL] select count(id) from e_enterprise limit 1000000;
影响的数据栏: 0
时间: 0.578ms
[SQL]
select count(*) from e_enterprise limit 1000000;
影响的数据栏: 0
时间: 0.532ms
----------------------------------------------------
[SQL] select count(*) from e_enterprise;
影响的数据栏: 0
时间: 0.532ms
[SQL]
select ROW_COUNT() from e_enterprise;
影响的数据栏: 0
时间: 1.282ms
--------------------------------------
[SQL] select count(1) from e_enterprise limit 1000000;
影响的数据栏: 0
时间: 0.532ms
[SQL]
select count(*) from e_enterprise limit 1000000;
影响的数据栏: 0
时间: 0.531ms
测试:500万条数据,表有聚集索引(id+时间字段),还有其它字段
count(*)与count(id):运行结果一样,速度几乎相同,均为13 "
count(*)与count(其他非空字段):非空字段无索引,结果相同,均为13 "
count(*)与count(其他有空的字段):有空字段无索引,前者13 ",后者1分多钟
可见:
若id不为空,count(*)与count(id)效率差不多
若id有空,count(*)比count(id)记录要多,但效率却要高,原因是count(*)直接统计记录总数,而count(id)还要判断id的值是否为空,因此降低了效率
下面是一些Mysql的用法(可能很少用)
1. 取得关于 information_schema的基本信息
查询所有的数据库 show databases;
2.use information_schema;
3.查询所有的表show tables;
4.超过1000行的表
select concat(table_schema,'.',table_name) as table_name,table_rows
from information_schema.tables where table_rows > 1000 order by table_rows desc;
5.没有主键的表
SELECT CONCAT(t.table_name,".",t.table_schema) as table_name
FROM information_schema.TABLES t
LEFT JOIN information_schema.TABLE_CONSTRAINTS tc
ON t.table_schema = tc.table_schema
AND t.table_name = tc.table_name
AND tc.constraint_type = 'PRIMARY KEY'
WHERE tc.constraint_name IS NULL
AND t.table_type = 'BASE TABLE';
下面是执行多个delete from 表 与 delete from 表 where id in (个数相同) 测试性能比较结果如下
Mysql5.1 window xp下
1:数据一千多左右
[SQL] delete from qa_questions where id = 100;
影响的数据栏: 0
时间: 0.015ms
[SQL]
delete from qa_questions where id = 101;
影响的数据栏: 1
时间: 0.032ms
[SQL]
delete from qa_questions where id = 102;
影响的数据栏: 0
时间: 0.015ms
[SQL]
delete from qa_questions where id = 103;
影响的数据栏: 1
时间: 0.016ms
[SQL]
delete from qa_questions where id = 104;
影响的数据栏: 1
时间: 0.015ms
[SQL]
delete from qa_questions where id = 105;
影响的数据栏: 1
时间: 0.015ms
[SQL]
delete from qa_questions where id = 106;
影响的数据栏: 1
时间: 0.016ms
[SQL]
delete from qa_questions where id = 107;
影响的数据栏: 1
时间: 0.016ms
[SQL]
delete from qa_questions where id = 108;
影响的数据栏: 1
时间: 0.015ms
[SQL]
delete from qa_questions where id = 109;
影响的数据栏: 1
时间: 0.015ms
[SQL]
delete from qa_questions where id in(
200,201,202,203,204,205,206,207,208,209
)
影响的数据栏: 9
时间: 0.016ms
2:数据量32万左右
[SQL] delete from e_template where id = 100;
影响的数据栏: 1
时间: 0.078ms
[SQL]
delete from e_template where id = 101;
影响的数据栏: 1
时间: 0.031ms
[SQL]
delete from e_template where id = 102;
影响的数据栏: 1
时间: 0.016ms
[SQL]
delete from e_template where id = 103;
影响的数据栏: 1
时间: 0.032ms
[SQL]
delete from e_template where id = 104;
影响的数据栏: 1
时间: 0.016ms
[SQL]
delete from e_template where id = 105;
影响的数据栏: 1
时间: 0.031ms
[SQL]
delete from e_template where id = 106;
影响的数据栏: 1
时间: 0.016ms
[SQL]
delete from e_template where id = 107;
影响的数据栏: 1
时间: 0.000ms
[SQL]
delete from e_template where id = 108;
影响的数据栏: 1
时间: 0.032ms
[SQL]
delete from e_template where id = 109;
影响的数据栏: 1
时间: 0.000ms
[SQL]
delete from e_template where id in(
200,201,202,203,204,205,206,207,208,209
)
影响的数据栏: 10
时间: 0.016ms
分析:从三十多万的数据可以看到 id = 100;时间: 0.078ms 而 where id in(
200,201,202,203,204,205,206,207,208,209) 只用了 0.016ms,此时有的where id = 是where id in(xx,xx,xx,....)消耗时间的4倍多
所有的删除加起来执行效率id=是in的 78+31+16+32+16+31+16+0+32+0/ 16倍,id=消耗的时间大约是where in的15倍多时间
3:我感觉数据量越大1000万以上越能体现 where id in 的优势。数据量大id = 反而没有id in(xx,xx,xx,...)删除多个性能好了。(其实数据1000多和三十二万的效率来讲 id = 差不多都是 id in的十几倍耗费时间,三十二万这种差距相比数据量1k又大多了。)