Oracle select distinct

做项目，写oracle查询语句的时候，总是喜欢滥用distinct关键字。但后来发现，某些场合下，distinct的性能开销非常大。

with 
diag_list as 
(select distinct diagnosis_desc from medrec.diagnosis where diagnosis_desc is not null),
diag_dict as 
(select distinct diagnosis_name,diagnosis_code from comm.diagnosis_dict where diagnosis_name is not null and diagnosis_code is not null)
select <span style="color:#ff0000;">distinct</span> t1.diagnosis_desc,t2.diagnosis_code,t2.diagnosis_name from diag_list t1,diag_dict t2 
where INSTR(t1.diagnosis_desc,t2.diagnosis_name)>0;

红色的distinct是完全没有必要的。而且，这个distinct的性能开销极大。

有时间的时候，可以使用profiler之类的工具对不同场景下的distinct对性能的影响进行量化分析。这里先做个备忘。