[救命]上月2小时的查询现在看来1个星期都结束不了了

askgyliu · 发表于 2007-6-25 23:05

THOMAS KYTE的意思是说SET PROCESSING 比 ROW PROCESSING快。RDBMS肯定得要尽量用SET PROCESSING。

dancingfire · 发表于 2007-6-27 18:42

看完众大虾们的讨论，才发现，路漫漫兮。。。

吾将上下而求索。。。

darkhorse_dh · 发表于 2007-7-2 23:11

今天终于看完了

hanjs · 发表于 2007-7-8 21:20

select /*+use_hash(a b) full(a) full(b) */
a.user_name,count(*) dup_num,
sum(least(a.stop_time,b.stop_time)-b.start_time) duptime_sum
from
adsl11_detail a,adsl11_detail b
where
a.user_name=b.user_name
and a.nas_ip=b.nas_ip
and a.nas_port=b.nas_port
and a.start_time<b.start_time and a.stop_time-1/288>b.start_time
and a.frame_ip!=b.frame_ip
group by
a.user_name
having
count(*)!=0;

这个sql改成这个好一点不？
[PHP]
select /*+use_hash(a b) full(a) full(b) */
a.user_name,count(*) dup_num,
sum(least(a.stop_time,b.stop_time)-b.start_time) duptime_sum
from
adsl11_detail a,adsl11_detail b
where
a.user_name=b.user_name
and a.nas_ip=b.nas_ip
and a.nas_port=b.nas_port
and a.start_time<b.start_time and a.stop_time>b.start_time+1/288
and a.frame_ip!=b.frame_ip
group by
a.user_name
having
count(*)>0;

/[/PHP]

看到各位大师的分析，学了不少！