mysql - 如何避免mysql存储过程中的IN(Select...)子查询

Question

我有一个存储函数，需要执行select ... from ... where foo IN (select ...)

问题是，与执行简单的 where = ... 相比，它慢得令人难以置信

由于已知这种语法很慢，是否有任何可能的方法将结果存储到变量中，然后在 IN () 条件中使用该变量？我不知道如何将多行查询存储到变量中。

这是使用变量的快速相等。 (但是错误的 b/c 它只执行 1 个值而不是多个值)

BEGIN

    DECLARE average DECIMAL(10,4);
    DECLARE skuAsin VARCHAR(30);

    SET skuAsin = (SELECT DISTINCT asin FROM inventory WHERE sku = aSku ORDER BY id DESC LIMIT 1);

    SET average = (
        SELECT  avg(unitsByDay) FROM (
            SELECT i.date, sum(units_ordered) as unitsByDay from inventory i
            WHERE
                i.asin = skuAsin &&
                i.marketplace_id = mid &&
                i.date between d1 and d2
            GROUP BY date
        ) as vel
    );

    RETURN average;

END;

这是使用IN(选择)的慢速

BEGIN

    DECLARE average DECIMAL(10,4);

    SET average = (
        SELECT  avg(unitsByDay) FROM (
            SELECT i.date, sum(units_ordered) as unitsByDay from inventory i
            WHERE
                i.asin IN (SELECT DISTINCT asin FROM inventory WHERE sku = aSku) &&
                i.marketplace_id = mid &&
                i.date between d1 and d2
            GROUP BY date
        ) as vel
    );

    RETURN average;

END;

Answer 1

而不是

i.asin IN (SELECT DISTINCT asin FROM inventory WHERE sku = aSku)

尝试使用EXISTS。

EXISTS (SELECT *
               FROM inventory ii
               WHERE ii.asin = i.asin
                     AND ii.sku = asku)

(我假设 asku 是一个变量，可能是过程的参数。如果没有使用正确的表别名来限定它。)

在库存(asin、sku) 上创建索引以支持它。

要考虑的另一个索引是外部查询的inventory (marketplace_id, date, asin)。