Question

这是工作的MySQL查询，是否可以将其转换为单个平面连接查询以提高性能？而不是使用下面的子查询。感谢

SELECT * from catAtt ca LEFT JOIN att a ON ca.attId = a.attId
WHERE ca.catId = 53 AND a.attCatId = 
(
SELECT DISTINCT eav.catId FROM ent LEFT JOIN eav ON ent.entId = eav.entId 
WHERE ent.key = 'somekey'
)

Answer 1

我将从简化查询开始。由于left join子句，where s都是内连接。此外，子查询中不应该需要select distinct。

select ca.*, att.*
from catAtt ca join
     att a 
     on ca.attId = a.attId 
where ca.catId = 53 and
      a.attCatId = (select eav.catId 
                    from ent join
                         eav 
                         on ent.entId = eav.entId 
                    where ent.key = 'somekey'
                   );

这建议使用索引：catAtt(catId, attId)，att(attId, attCatId)，ent(key, entId)和eav(entId, catId)。

您还可以将where中的子查询移动到from。我认为这不会对性能产生太大影响。它不相关，因此它运行一次，并返回一个值。

Answer 2

let disposable = Observable.combineLatest(objectA.asObservable().skip(1), 
                                          objectB.asObservable().skip(1))
                           .throttle(30, 
                                     scheduler: ConcurrentDispatchQueueScheduler(qos: .background))
                           .subscribe(onNext: { [unowned self] _ in
                    self.update()
            })

尝试以上代码。

希望这会对你有所帮助。

Answer 3

我在sql server上尝试了三种类型的查询。

SELECT * from catAtt ca JOIN att a ON ca.attId = a.attId
WHERE ca.catId = 53 AND a.attCatId = 
(
SELECT DISTINCT eav.catId FROM ent LEFT JOIN eav ON ent.entId = eav.entId 
WHERE ent.key = 'somekey'
)

SELECT * from catAtt ca 
JOIN att a ON ca.attId = a.attId
JOIN 
(
  SELECT DISTINCT eav.catId FROM ent 
  LEFT JOIN eav ON ent.entId = eav.entId 
  WHERE ent.key = 'somekey'
)Z ON a.attCatId = Z.catId 
WHERE ca.catId = 53 

SELECT * from catAtt ca JOIN att a ON ca.attId = a.attId
WHERE ca.catId = 53 AND 
EXISTS 
   (
    SELECT 1 FROM ent LEFT JOIN eav ON ent.entId = eav.entId 
    WHERE ent.key = 'somekey' AND a.attCatId = eav.catId
   )

每个查询的查询成本相同，为33％。

猜猜数据库比我们更聪明。

将子查询转换为连接查询

3 个答案: