oracle优化技巧及实例（总结）

1.关于exists和in

in是循环的方式，在内存中处理，

exists是执行数据库查询，

select tpd.personaccountid,sum(nvl(tpd.CREDIT_SUM, 0)) as bjsr, sum(nvl(tpd.INTEREST_INCOME, 0)) as bjsr, sum(nvl(tpd.DEBIT_SUM, 0)) as bjzc from TGP_PERSON_DETAIL tpd
left join TAP_FUNDBUSINESS tfb
on tpd.FUNDBUSINESS_ID = tfb.FUNDBUSINESS_ID
where tpd.PERSONACCOUNTID
in (
select personaccountid from TGP_PERSONACCOUNT pa
left join tgp_person person
on pa.personid = person.personid
where person.zjhm = '330903196209160251'
) group by tpd.personaccountid

执行100次耗时0.017S左右

select tpd.personaccountid,sum(nvl(tpd.CREDIT_SUM, 0)) as bjsr, sum(nvl(tpd.INTEREST_INCOME, 0)) as bjsr, sum(nvl(tpd.DEBIT_SUM, 0)) as bjzc from TGP_PERSON_DETAIL tpd
left join TAP_FUNDBUSINESS tfb
on tpd.FUNDBUSINESS_ID = tfb.FUNDBUSINESS_ID
where
exists (
select personaccountid from (select personaccountid from TGP_PERSONACCOUNT pa
left join tgp_person person
on pa.personid = person.personid
where person.zjhm = '330903196209160251') a
where tpd.personaccountid = a.personaccountid
) group by tpd.personaccountid

执行100次耗时0.020S左右

以上都为SELECT FORM a IN b(OR EXISTS B) B中数据少的时候采用IN 多的时候采用EXISTS

select tpd.personaccountid,sum(nvl(tpd.CREDIT_SUM, 0)) as bjsr, sum(nvl(tpd.INTEREST_INCOME, 0)) as bjsr, sum(nvl(tpd.DEBIT_SUM, 0)) as bjzc from TGP_PERSON_DETAIL tpd
left join TAP_FUNDBUSINESS tfb
on tpd.FUNDBUSINESS_ID = tfb.FUNDBUSINESS_ID
HAVING
exists (
select personaccountid from (select personaccountid from TGP_PERSONACCOUNT pa
left join tgp_person person
on pa.personid = person.personid
where person.zjhm = '330903196209160251') a
where tpd.personaccountid = a.personaccountid
) group by tpd.personaccountid

执行100次耗时2.36S左右

因此先筛选数据再分组效率更高