oracle中如何让exists中的group by 更快些

Question

像这样在exists中又写了一个group by，但是很慢，谁能给优化下或者改进下。
select * from tcf_bill_after_t d
where perserve03 is null
and exists(
select a.account_idnew
from
(
select FEE_KIND,ACCOUNT_ID,USER_ID,sum(fee1+fee2+fee3+fee4+fee11+fee12+fee13+fee14) fee
from bf_bill_t_before
group by ACCOUNT_ID,USER_ID,FEE_KIND
) a
where a.fee<0
and d.PERSERVE01 ='old'
and a.account_id = d.account_idnew
and a.perserve02='0'

)
and table_source=1
and city_code in('0411','0427')
;

Answer 1

这样试试 别用exists了

select d.* from tcf_bill_after_t d,(
select a.account_idnew
from
(
select FEE_KIND,ACCOUNT_ID,USER_ID,sum(fee1+fee2+fee3+fee4+fee11+fee12+fee13+fee14) fee
from bf_bill_t_before
group by ACCOUNT_ID,USER_ID,FEE_KIND
) a
where a.fee<0
and d.PERSERVE01 ='old'
and a.account_id = d.account_idnew
and a.perserve02='0'
) x

where perserve03 is null and d.account_id = x.account_idnew
and existsand table_source=1
and city_code in('0411','0427')
;

Answer 2

加几个索引吧，和exists关系不大

Answer 3

视乎没有必要求和吧 exist 是判断存在而已