sql - 我怎样才能避免额外的内部连接来优化这个查询？

Question

有什么方法可以避免内部连接以从客户表中获取客户名字和姓氏，以及如何优化执行计划？

SELECT c1.firstname, c1.lastname, t.*
FROM customer c1
INNER JOIN
    (select c.Customerid ,count(si.CustomerID) as 'No Of Orders'
    from Customer c
    LEFT OUTER join SalesInvoice si ON si.CustomerID = c.CustomerID
    GROUP by c.CustomerID) as t
ON c1.CustomerID = t.Customerid

Answer 1

可能有两种方法可以做到这一点。一种方法是将所有非聚合列包含到分组列表中，如下所示：

SELECT
    c.firstname,
    c.lastname,
    c.CustomerId,
    count(si.CustomerID) as 'No Of Orders'
FROM
    customer c
LEFT OUTER JOIN
    SalesInvoice si ON si.CustomerID = c.CustomerID
GROUP BY
    c.CustomerID,
    c.firstname,
    c.lastname

另一种方法是在选择列表中使用聚合函数（应该使用最小值或最大值）：

SELECT
    MIN(c.firstname) as firstname,
    MIN(c.lastname) as lastname,
    c.CustomerId,
    count(si.CustomerID) as 'No Of Orders'
FROM
    customer c
LEFT OUTER JOIN
    SalesInvoice si ON si.CustomerID = c.CustomerID
GROUP BY
    c.CustomerID

我认为第一个查询的性能可能更好。为了提高整体性能，您需要创建一个包含或覆盖索引。

您可以查看以下链接，了解包含的索引：

1. 为什么在创建索引时使用 INCLUDE 子句？
2. http://msdn.microsoft.com/en-IN/library/ms190806.aspx

Answer 2

使用OVER()子句尝试此选项

SELECT c1.Customerid, c1.firstname, c1.lastname, 
       COUNT(si.CustomerID) OVER(PARTITION BY c1.CustomerID) AS 'No Of Orders'
FROM customer c1 LEFT JOIN SalesInvoice si ON si.CustomerID = c1.CustomerID

为了提高性能，您需要此索引

CREATE INDEX ix_Customerid_Customer ON Customer(Customerid) INCLUDE(firstname, lastname)

Answer 3

试试那个：

SELECT
    c.firstname,
    c.lastname,
    c.CustomerId,
    count(si.CustomerID) as 'No Of Orders'
FROM
    customer c
LEFT OUTER JOIN
    SalesInvoice si ON si.CustomerID = c.CustomerID
GROUP BY
    c.CustomerID,
    c.firstname,
    c.lastname