原文:
乙個sql的優化
目的:為了查詢某天某個伺服器上的登入id的個數
剛開始編寫的sql:
select count(a.mac) logusers from log_maclogin_all a
where ismoni != 1
and logintime <= '2015-02-01 23:59:59' and logintime >= '2015-02-01 00:00:00'
and a.mac in (select mac from inst_user_mac b
where dotime <= '2015-01-30 23:59:59'
and dotime >= '2015-01-30 00:00:00 and serverkey='p1s1' );
執行時間為:33s
對於這個sql，首先的優化點在於"in",因為測試發現in條件裡面的資料差不多將近萬以上的資料..
第一次優化：把in改為exists後，效率有一點的提公升
select count(mac) logusers from log_maclogin_all a
where ismoni != 1
and logintime <= '2015-02-01 23:59:59' and logintime >= '2015-02-01 00:00:00'
and exists (select mac from inst_user_mac b
where dotime <= '2015-01-30 23:59:59'
and dotime >= '2015-01-30 00:00:00 and serverkey='p1s1'
and a.mac = b.mac );
執行時間為：26s
第二次優化：在網上查了下，有的說用join會快一些。把exists改為join試試
select count(a.mac) logusers from log_maclogin_all a inner join
(select mac from inst_user_mac where dotime <= '2015-01-30 23:59:59'
and dotime >= '2015-01-30 00:00:00
and serverkey='p1s1' ) b
on a.mac = b.mac
where a.ismoni != 1
and a.logintime <= '2015-02-01 23:59:59' and a.logintime >= '2015-02-01 00:00:00';
執行時間為2.6s，效能有了明顯的提高啊
第三次優化:把》= <= 改為between and 這樣會減少把資料查出來後的計算操作
select count(a.mac) logusers from log_maclogin_all a inner join
(select mac from inst_user_mac where dotime between '2015-01-30 00:00:00
and '2015-01-30 23:59:59'
and serverkey='p1s1' ) b
on a.mac = b.mac
where a.ismoni != 1
and a.logintime between '2015-02-01 00:00:00'
and '2015-02-01 23:59:59';
執行時間為2.4s，提公升了0.2s
第四次優化:假如要查某一天的資料不如直接date_format();
select count(a.mac) logusers from log_maclogin_all a inner join
(select mac from inst_user_mac where date_format(dotime,'%y%m%d')='20150130' and serverkey='p1s1' ) b
on a.mac = b.mac
where a.ismoni != 1 and date_format(a.logintime,'%y%m%d') = '20150201';
執行時間為2.36s,又提公升了一點點..
關於乙個加法優化的乙個地方
include include include base.h int main int argc,char argv,char envp 下面是彙編 01291000 55 push ebp 01291001 8bec mov ebp,esp 01291003 56 push esi 0129100...
乙個sql的例子
select dbo.userinfo.username,dbo.userinfo.usertruename,dbo.userinfo.useremail,dbo.userinfo.usermobile,dbo.userinfo.usertelephone,dbo.userinfo.usercar,...
乙個複雜的sql
select f.course node info id as nodeid,c.course node name as nodename,c.course node type as nodelevel,c.course code,case when select course node info ...