使用计数优化SQL查询

Question

我使用ASP.NET成员资格，最重要的是我添加了一些与用户有一对多关系的表。

一个用户有很多笔，笔记本等

我编码一个自定义的成员提供和已经修改了选择查询检索所有关于用户以及用户具有笔的数量和信息“工具”的总数，但查询显然是灾难性的。 无论如何得到这些信息？ 这是我曾尝试：

SET STATISTICS TIME ON 
SELECT m.Email, m.PasswordQuestion, m.IsApproved, m.CreateDate, m.LastLoginDate, 
     u.LastActivityDate, m.LastPasswordChangedDate, u.UserId, m.IsLockedOut, 
     m.LastLockoutDate, CAST(m.Comment as nvarchar(max)) as Comment, 
     COUNT(DISTINCT cl.Id) AS NumberOfPens, 
     COUNT(DISTINCT cf.Id)+ 
      COUNT(DISTINCT cm.Id) + 
      COUNT(DISTINCT ce.Id) + 
      COUNT(DISTINCT cl.Id) + 
      COUNT(DISTINCT cp.Id) + 
      COUNT(DISTINCT csl.Id)+ 
      COUNT(DISTINCT cs.Id) + 
      COUNT(DISTINCT cshl.Id) AS NumberOfTools 
FROM aspnet_Membership AS m 
     INNER JOIN aspnet_Users AS u ON m.UserId = u.UserId 
     INNER JOIN Pens AS cf ON u.UserId = cf.Owner 
     INNER JOIN Notebooks AS cm ON u.UserId = cm.Owner 
     INNER JOIN Rulers AS ce ON u.UserId = ce.Owner 
     INNER JOIN Calculators AS cl ON u.UserId = cl.Owner 
     CROSS JOIN aspnet_Applications AS a 
GROUP BY u.UserId, m.Email, m.PasswordQuestion, m.IsApproved, 
     m.CreateDate, m.LastLoginDate, u.LastActivityDate, 
     m.LastPasswordChangedDate, m.IsLockedOut, m.LastLockoutDate, 
     CAST(m.Comment as nvarchar(max)) 

SET STATISTICS TIME OFF 

9分钟11秒91级的用户！

Answer 1

作为一个快速打击，最终拿出cross join - 你没有使用它，它只是增加你的结果并放慢速度。

此外，你可以做接合前的计数，这将节省group by并允许RDBMS并行查询，像这样：

SELECT  
    m.Email, 
    m.PasswordQuestion, 
    m.IsApproved, 
    m.CreateDate, 
    m.LastLoginDate, 
    u.LastActivityDate, 
    m.LastPasswordChangedDate, 
    u.UserId, 
    m.IsLockedOut, 
    m.LastLockoutDate, 
    CAST(m.Comment as nvarchar(max)) as Comment, 
    PenCount AS NumberOfPens, 
    PenCount + NotebookCount + RulerCount + CalculatorCount AS NumberOfTools 
FROM 
    aspnet_Membership AS m 
    INNER JOIN aspnet_Users AS u ON 
     m.UserId = u.UserId 
    INNER JOIN (select owner, count(1) as PenCount from Pens group by owner) AS cf ON 
     u.UserId = cf.Owner 
    INNER JOIN (select owner, count(1) as NotebookCount from Notebooks group by owner) AS cm ON 
     u.UserId = cm.Owner 
    INNER JOIN (select owner, count(1) as RulerCount from Rulers group by owner) AS ce ON 
     u.UserId = ce.Owner 
    INNER JOIN (select owner, count(1) as CalculatorCount from Calculators group by owner) AS cl ON 
     u.UserId = cl.Owner 

需要注意的是，如果一个用户可能没有所有工具，你会想要做LEFT连接代替inner连接，然后将所有*Count列与coalesce(Count, 0)包装在一起。

Answer 2

SELECT m.Email, m.PasswordQuestion, m.IsApproved, m.CreateDate, m.LastLoginDate, 
     u.LastActivityDate, m.LastPasswordChangedDate, u.UserId, m.IsLockedOut, 
     m.LastLockoutDate, CAST(m.Comment as nvarchar(max)) as Comment, 
     penCount, 
     notebookCount + rulerCount + calculatorCount AS toolCount 
FROM aspnet_Membership m 
JOIN aspnet_Users u 
ON  u.userId = m.userId 
CROSS APPLY 
     (
     SELECT COUNT(*) 
     FROM pens 
     WHERE owner = u.UserId 
     ) p (penCount) 
CROSS APPLY 
     (
     SELECT COUNT(*) 
     FROM notebooks 
     WHERE owner = u.UserId 
     ) n (notebookCount) 
CROSS APPLY 
     (
     SELECT COUNT(*) 
     FROM rulers 
     WHERE owner = u.UserId 
     ) r (rulerCount) 
CROSS APPLY 
     (
     SELECT COUNT(*) 
     FROM calculators 
     WHERE owner = u.UserId 
     ) c (calculatorCount)