Question

我有一个产品和产品表的产品销售以及在这些产品上完成的所有销售操作。这些表通过Sales.PRODUCT_ID列连接。

我想今天购买10种最常销售的产品，我所做的就是：

SELECT product.* , COUNT( sale.ID ) SUMSELL
FROM Products product
LEFT JOIN 
   Sales sale 
   ON 
       sale.ANKIETA_ID = product.ID AND 
       sale.DATE >= DATE_SUB( NOW( ) , INTERVAL 1 DAY )
GROUP BY product.ID
ORDER BY SUMSELL DESC
LIMIT 0 , 10

但它的表现非常缓慢。如何提高此特定查询的效果？

Answer 1

摆脱LEFT JOIN：

SELECT  product.*, sumsell
FROM    (
        SELECT  ankieta_id, COUNT(*) AS sumsell
        FROM    sale
        WHERE   sale.DATE >= DATE_SUB( NOW( ) , INTERVAL 1 DAY )
        GROUP BY
                ankieta_id
        ) s
JOIN    product
ON      product.id = s.ankieta_id
ORDER BY
        SUMSELL DESC
LIMIT 10

在sale (date, ankieta_id)上建立一个索引，以便快速运作。

如果您的桌子可能销售的产品少于10，但您需要展示10，请执行以下操作：

SELECT  *
FROM    (
        SELECT  product.*, sumsell
        FROM    (
                SELECT  ankieta_id, COUNT(*) AS sumsell
                FROM    sale
                WHERE   sale.DATE >= DATE_SUB( NOW( ) , INTERVAL 1 DAY )
                GROUP BY
                        ankieta_id
                ) s
        JOIN    product
        ON      product.id = s.ankieta_id
        ORDER BY
                SUMSELL DESC
        LIMIT 10
        ) p
UNION ALL
SELECT  *
FROM    (
        SELECT  product_id, 0
        FROM    products p
        WHERE   products NOT IN
                (
                SELECT  *
                FROM    sale
                WHERE   ankieta_id = p.id
                        AND sale.DATE >= DATE_SUB( NOW( ) , INTERVAL 1 DAY )
                ) p2
        LIMIT 10
        ) p2
ORDER BY
        sumsell DESC
LIMIT 10

Answer 2

您在左连接中使用sale.DATE >= DATE_SUB( NOW( ) , INTERVAL 1 DAY )条件。我猜这就是为什么它很慢。

我建议：

将日期条件移至WHERE子句
使用JOIN代替JOIN
在Sales.DATE上添加和索引也有助于缩小查询范围

SELECT product.* , COUNT( sale.ID ) SUMSELL
FROM Products product
JOIN Sales sale 
  ON sale.ANKIETA_ID = product.ID
WHERE sale.DATE >= DATE_SUB( NOW( ) , INTERVAL 1 DAY )
GROUP BY product.ID
ORDER BY SUMSELL DESC
LIMIT 0 , 10

如何增加最后一天计数查询性能

2 个答案: