select count 时 *速度快还是ID速度快

[whb2597758]

select count(*) from list 还是 select count(ID) from list

这两天查询语句哪个速度更快？ list表大概有800多万行数据， ID是主键！

我试了一下，感觉没什么区别！ 请各位大虾，亲们解答！

统计时，牵扯的表多了，有时需要10多秒才能执行完毕，纠结啊。。。！ 必须优化！


附上存储过程，各位大虾看看有什么可以优化的


USE [MetenSite]
GO
/****** 对象: StoredProcedure [dbo].[GetCurrentMonthRegisterNum] 脚本日期: 06/22/2011 10:15:57 ******/
SET ANSI_NULLS ON
GO
SET QUOTED_IDENTIFIER ON
GO
CREATE proc [dbo].[GetCurrentMonthRegisterNum](@beginTime dateTime,@endTime dateTime, @ids nvarchar(1000))
as


IF @ids IS NOT NULL
BEGIN

select
(select count(*) from dbo.Member where registerTime BETWEEN @beginTime AND @endTime AND [SupportID] IN (SELECT IDS FROM f_split(@ids, ',') )) as monthRegisterUser,
(select count(*) from dbo.CustomerList where ViewDate BETWEEN @beginTime AND @endTime AND [TID] IN (SELECT IDS FROM f_split(@ids, ',') )) as monthIp,
(select Sum(pv) from dbo.CustomerList where ViewDate BETWEEN @beginTime AND @endTime AND [TID] IN (SELECT IDS FROM f_split(@ids, ',') )) as monthPV,
(select count(*) from dbo.Member where supportId<>2 and registerTime BETWEEN @beginTime AND @endTime AND [SupportID] IN (SELECT IDS FROM f_split(@ids, ',') )) as monthSupportRegisterUser,
(select count(*) from dbo.CustomerList where TID !=-1 and ViewDate BETWEEN @beginTime AND @endTime AND [TID] IN (SELECT IDS FROM f_split(@ids, ',') )) as monthSupportIp,
(select sum(pv) from dbo.CustomerList where TID !=-1 and ViewDate BETWEEN @beginTime AND @endTime AND [TID] IN (SELECT IDS FROM f_split(@ids, ',') )) as monthSupportPV


END

ELSE
BEGIN

select
(select count(*) from dbo.Member where registerTime BETWEEN @beginTime AND @endTime) as monthRegisterUser,
(select count(*) from dbo.CustomerList where ViewDate BETWEEN @beginTime AND @endTime) as monthIp,
(select Sum(pv) from dbo.CustomerList where ViewDate BETWEEN @beginTime AND @endTime) as monthPV,
(select count(*) from dbo.Member where supportId<>2 and registerTime BETWEEN @beginTime AND @endTime ) as monthSupportRegisterUser,
(select count(*) from dbo.CustomerList where TID !=-1 and ViewDate BETWEEN @beginTime AND @endTime) as monthSupportIp,
(select sum(pv) from dbo.CustomerList where TID !=-1 and ViewDate BETWEEN @beginTime AND @endTime) as monthSupportPV

END

[chenhongjun0624]

应该没有差别

[lijian8552]

性能差不多的！

[zy35147972122]

像你这样只统计行数的话不需要用COUNT(*)或者COUNT(ID)的
直接用count（1）不更好

[-晴天]

速度上没什么太大的差别

[xuexiaodong2009]

记得看过说是这个快select count(ID) from list

[mingpei0703]

优化，学习，友情帮顶

[唐诗三百首]

10多秒才执行完毕,不排除是锁争的问题.可考虑降低事务隔离级别,

  select count(ID) from list (nolock)

[GoAwayZ]

count(*),count(id)一样

[htl258_Tony]

几乎没差别，SQL优化器会自动找索引列去聚合

[叶子]

速度上差不多，你可以弄个大数量的测试一下。

[rucypli]

你看一下执行计划啊

[水向东流时光怎么偷]

一样

[zyxlanmo]

速度一样的 不区分的

[qgqch2008]

一樣！

[BigInt]

要抓关键，你的语句的关键是语句本身是嵌套查询，这里有很大优化空间

如果语句不修改，那关键是这些日期字段是否有索引

[BigInt]

把(SELECT IDS FROM f_split(@ids, ',') 的内容放到一个临时表或者表变量，至少函数不必调用多次，这些语句的优化不做只能说明编写者根本没有优化意识，而这时候去探究count(*)、count(id)、count(1)这些微小的性能差别，窃以为不可取。

[--小F--]

查询分析器会自动调整

[q107770540]

select count(1) from list

[wbj2009]

一般情况下，Select Count (*)和Select Count(1)两着返回结果是一样的

假如表沒有主键(Primary key), 那么count(1)比count(*)快，

如果有主键的話，那主键作为count的条件时候count(主键)最快

如果你的表只有一个字段的话那count(*)就是最快的

count(*) 跟 count(1) 的结果一样，都包括对NULL的统计，而count(column) 是不包括NULL的统计



1、select 1 与 select *的区别
selelct 常量 from ... 对应所有行，返回的永远只有一个值，即常量 。所以正常只会用来判断是否有还是没有（比如exists子句）。而select * from ... 是返回所有行的所有列。
性能上的差异，关键看你的from和where子句。比如说如果你的where条件中可以通过索引，那显然 select 1 from ... 的性能比 select * from ... 好。
2、select sum(1)的使用
select count(*)返回所有满足条件的记录数，此时同select sum(1)
但是sum()可以传任意数字，负数、浮点数都可以，返回的值是传入值n*满足条件记录数m