Question

我正在为现有应用程序编写附加内容，需要获取各种事务类型的数量总和。每种交易类型都有一组唯一的标准ta_code，ta_type和ta_asc

下面的查询确实返回了必要的信息，但最终，反式表有大约500万条记录，所以运行需要4分钟

SELECT t.qty AS tQty, sQty, rQty, mQty, lQty, aQty, xQty, t.itmcode
FROM trans AS t
LEFT JOIN 
    (
        SELECT sum(qty) as sQty, itmcode
        FROM trans
        WHERE 
            site = 127 
            AND ta_code = '12'
        GROUP BY itmcode
    )as tblS ON tblS.itmcode = t.itmcode
LEFT JOIN 
    (
        SELECT sum(qty) as rQty, itmcode
        FROM trans
        WHERE 
            site = 127 
            AND ta_code = '12' AND ta_type = '1'
        GROUP BY itmcode
    )as tblR ON tblR.itmcode = t.itmcode
LEFT JOIN 
    (
        SELECT sum(qty) as mQty, itmcode
        FROM trans
        WHERE 
            site = 127 
            AND ta_code = '15' AND ta_type = '4'
        GROUP BY itmcode
    )as tblM ON tblM.itmcode = t.itmcode
LEFT JOIN 
    (
        SELECT sum(qty) as lQty, itmcode
        FROM trans
        WHERE 
            site = 127 
            AND ta_asc = '147'
        GROUP BY itmcode
    )as tblL ON tblL.itmcode = t.itmcode
LEFT JOIN 
    (
        SELECT sum(qty) as aQty, itmcode
        FROM trans
        WHERE 
            site = 127 
            AND ta_asc = '155'
        GROUP BY itmcode
    )as tblA ON tblA.itmcode = t.itmcode
LEFT JOIN 
    (
        SELECT sum(qty) as xQty, itmcode
        FROM trans
        WHERE 
            site = 127 
            AND ta_asc = '155' AND ta_code = '15' AND ta_type = '4'
        GROUP BY itmcode
    )as tblX ON tblX.itmcode = t.itmcode                
INNER JOIN item AS i ON i.code = t.itmcode       
WHERE t.site = 127
GROUP BY itmcode

每个连接似乎都会增加效率，但即使单独运行每个连接选择也需要10 +秒才能运行。我们只对该数据库具有读访问权限，因此无法以任何方式对其进行更改。

如果有人能够建议能够更有效地实现这一目标的方法或结构，我将不胜感激。

Answer 1

以下是我要尝试的一些事情：

在itmCode列上添加索引
移动INNER JOIN项目AS i ON i.code = t.itmcode作为第一次加入
使用临时表“SELECT sum（qty）as sQty，itmcode FROM trans WHERE site = 127 AND ta_code = '12' GROUP BY itmcode“然后在其他左连接中使用它来添加ta_type，因为你在每个左连接中不必要地重复这个查询。

如果您使用MS SQL，我也总是建议学习和使用MS SQL配置文件。

Answer 2

这可能值得一试：删除所有“左连接”子句，并计算条件总和，如下所示：

    SELECT 
    ISNULL(SUM(CASE WHEN ta_code = '12' and ta_type='1' THEN Qty ELSE 0 END), 0) as rQty,
    ISNULL(SUM(CASE WHEN ta_code = '15' and ta_type='4' THEN Qty ELSE 0 END), 0) as mQty

    FROM trans AS t

    WHERE t.Site = 127
    GROUP BY itmcode

优化大表上的多连接查询

2 个答案: