sql的性能优化之——distinct与group by

1. 表A （uid,bid) ，
   uid代表：用户id
   bid代表：uid关注的用户id
   表数据示例：
   uid bid

---

         1 2
         2 1
         1 3

------我的答案

select
    t1.uid,
    t1.bid
from tbl t1
join
   (select
        uid,bid
   from tbl)t2
on t1.uid = t2.bid and t1.bid = t2.uid

实际上，一需要考虑到数据中的去重问题

chatgpt提供

使用distinct

1.对数据去重

distinct uid,bid

2.再加一个顺序排序

order by bid,uid

3.综合sql

SELECT DISTINCT t1.uid, t1.bid
FROM tab t1
JOIN tableA t2 ON t1.uid = t2.bid AND t1.bid = t2.uid
ORDER BY t1.uid, t1.bid;

-----对于性能，考虑到使用group by 

1.使用union

select
    uid,bid
from tbl 
union
select
    bid,uid
from tbl

2.进行筛选过滤数据

select
    uid,bid
from tbl 
where (uid,bid) in ()t1

3.去重分组

select
    uid,bid,
    count(*) 
from t2
group by uid,bid

4.过滤数据为2

select
    uid,bid
from tbl
having count(*) = 2

5.综合sql

SELECT uid, bid
FROM tableA
WHERE (uid, bid) IN (
    SELECT uid, bid
    FROM tableA
    UNION
    SELECT bid, uid
    FROM tableA
)
GROUP BY uid, bid
HAVING COUNT(*) = 2;

--- 发现having用法

意外发现，使用group by 的时候，聚合可以直接在having后面完成，无需写在select这里。