如何使用 sql server exists 子句进行优化？

Posted 2023-03-28

【中文标题】如何使用 sql server exists 子句进行优化？

【英文标题】：How to do optimization with sql server exists clause?

【发布时间】：2016-07-28 06:51:53

【问题描述】：

我有一个包含 exists 子句的 sql 查询。以前在两个表之间存在内部联接，但由于一些新条件，我用存在子句替换了内部联接。以下是我的查询-

SELECT id, 
       Count(id) AS 'ItemCount' 
FROM   table1 t1 
WHERE  ( EXISTS(SELECT DISTINCT messageid 
                FROM   table2 t2 
                WHERE  t2.messageid = t1.chatmessageid 
                       AND t2.receiverid = 'xyzo-1234' 
                       AND t2.groupid = 'wxfx-1234' 
                       AND ( t1.userid = 'ttew-3457' 
                              OR t1.userid IS NULL )) 
          OR ( t1.userid = 'ttew-3457' ) ) 
       AND ( t1.machineid != 'C34F-A609' ) 
       AND ( t1.createdtime > '2016-07-20 :: 07:21:51.9118534' ) 
       AND ( t1.createdtime <= '2016-07-27 :: 13:25:30.1016155' ) 
       AND ( t1.groupid = 'wxfx-1234' ) 
GROUP  BY t1.id 
ORDER  BY t1.id 

但是这个查询比上一个查询花费了更多的时间，我想进一步优化它。谁能建议我如何进一步优化此查询？

【问题讨论】：

好吧，对于初学者来说，DISTINCT 在 EXISTS 中毫无意义，因为它只关心 0 行和多于 0 行之间的区别。

EXISTS(...OR t1.UserId IS NULL - 你确定吗？

如果你有可以有 NULL 值的列，用类似isnull(t1.UserId, '')的东西重写

我建议您开始使用DATETIME2 数据类型来存储日期+时间信息。这将使比较/查找更快。

【参考方案1】：

SELECT Id, COUNT(Id) as 'ItemCount' 
FROM table1 t1 WHERE (
                    EXISTS( 
                            ***SELECT Count 1 0 FROM table2 t2***  
                            WHERE t2.MessageId = t1.ChatMessageId AND t2.ReceiverId ='xyzo-1234' 
                            AND t2.GroupId='wxfx-1234'  
                            AND (t1.UserId ='ttew-3457' OR t1.UserId IS NULL) 
                            ) 
                OR (t1.UserId ='ttew-3457')
            ) 
        AND  (t1.MachineId != 'C34F-A609') AND (t1.CreatedTime > '2016-07-20 :: 07:21:51.9118534')
        AND (t1.CreatedTime <= '2016-07-27 :: 13:25:30.1016155') AND (t1.GroupId='wxfx-1234') 
GROUP BY t1.Id  
ORDER BY t1.Id