每次我选择MySQL函数时，它的执行时间都会增加

有点奇怪。我有一个函数一开始运行得很好，比如说，当选择返回值加上其他一些列时，执行15毫秒，获取10毫秒。但是，如果不断刷新同一个查询，查询的执行就会上升。所以先是15秒，然后是17秒，然后。。。我一路跑到900米。这主要是随着时间的推移而上升的获取，但执行也是如此。因此，在最后，它将是600毫秒的获取和300毫秒的执行。你知道发生了什么事吗？

函数。我只尝试了一个简单的IF/ELSEIF，但它在性能方面给出了完全相同的结果。

create function get_table(var_account_id int unsigned) returns varchar(20)
    reads sql data
BEGIN
    RETURN IF(
            (SELECT EXISTS(SELECT TRUE
                           FROM TableA
                           WHERE account_id = var_account_id
                             AND expiring_at > CURRENT_TIMESTAMP)), 'TableA',
            IF((SELECT EXISTS(SELECT TRUE
                              FROM TableB
                              WHERE account_id = var_account_id
                                AND expiring_at > CURRENT_TIMESTAMP)), 'TableB',
               IF((SELECT EXISTS(SELECT TRUE
                                 FROM TableC
                                 WHERE account_id = var_account_id
                                   AND expiring_at > CURRENT_TIMESTAMP)), 'TableC',
                  IF((SELECT EXISTS(SELECT TRUE
                                    FROM TableD
                                    WHERE account_id = var_account_id
                                      AND expiring_at > CURRENT_TIMESTAMP)), 'TableD',
                     NULL)
                   )));
END;

用var_account_id=1 运行一次后解释函数

9,SUBQUERY,TableD,,ref,"TableD_expiring_at_index,TableD_account_id_index",TableD_account_id_index,4,const,1,100,Using where
7,SUBQUERY,TableC,,ref,"TableC_account_id_index,TableC_expiring_at_index",TableC_account_id_index,4,const,1,5,Using where
5,SUBQUERY,TableB,,ref,"TableB_expiring_at_index,TableB_account_id_index",TableB_account_id_index,4,const,1,9.26,Using where
3,SUBQUERY,TableA,,ref,"TableA_expiring_at_index,TableA_account_id_index",TableA_account_id_index,4,const,1,100,Using where

在account_id和expiring_at上设置复合索引对没有任何影响

我运行一个类似的简单查询

SELECT TableXYZ.*, get_table(TableXYZ.account_id) AS some_value FROM TableXYZ LIMIT 500;

我在更复杂的查询中运行过它，但结果总是一样的，一开始很快，重新运行同一个SELECT后很慢，比如说连续30秒每秒5次。即使在我让MySQL冷却一段时间后，回来，第一次运行仍然是900ms。我相信它会继续上升。解决这个问题的唯一方法是在windows中重新启动mysql服务。

解释SELECT：

1,SIMPLE,TableXYZ,,ALL,,,,,695598,100,

如果重要的话，我会在Windows 10上运行这些，本地的。

SELECT X.tabName
FROM (
SELECT 'TableA' as tabName, 1 as tabNr, account_id, expiring_at FROM TableA WHERE account_id = var_account_id AND expiring_at > CURRENT_TIMESTAMP
UNION
SELECT 'TableB' as tabName, 2 as tabNr, account_id, expiring_at FROM TableB WHERE account_id = var_account_id AND expiring_at > CURRENT_TIMESTAMP
UNION
...
    ) X ORDER BY tabNr LIMIT 1

DECLARE tabName VARCHAR(50);
SET tabName := SELECT 'TableA' FROM TableA WHERE account_id = var_account_id AND expiring_at > CURRENT_TIMESTAMP;
IF (tabName IS NULL) THEN
   SET tabName := SELECT 'TableB' FROM TableB WHERE account_id = var_account_id AND expiring_at > CURRENT_TIMESTAMP;
END IF;
and so on...
RETURN tabName;

RETURN COALESCE(
    IF (EXISTS(...), "A", NULL),
    IF (EXISTS(...), "B", NULL),
    IF (EXISTS(...), "C", NULL),
    IF (EXISTS(...), "D", NULL),
    IF (EXISTS(...), "E", NULL)
               )

SELECT 1
    FROM TableA
    WHERE account_id = var_account_id
      AND expiring_at > CURRENT_TIMESTAMP