使用计算的 where 子句的 MySQL 查询性能

Question

使用MySQL，1M记录表，存在索引“some_timestamp”。

WHERE 子句使用计算，以便可以对不同时间段的记录进行计数。

需要很长时间才能执行。

有更好、更快的方法吗？？？

这是当前代码:

set @now = now();

(SELECT 
    "45 Days" as "interval", count(*) as "thecount"
    FROM blah.yada
    where TIMESTAMPDIFF(day,some_timestamp,@now) <= 45
        )

UNION
(SELECT 
    "90 Days" as "interval", count(*) as "thecount"
    FROM blah.yada
    where TIMESTAMPDIFF(day,some_timestamp,@now) > 45 and TIMESTAMPDIFF(day,event_timestamp,@now) <=90
        )

UNION
(SELECT 
    "the rest" as "interval", count(*) as "thecount"
    FROM blah.yada
    where TIMESTAMPDIFF(day,some_timestamp,@now) > 90 
        )

Answer 1

在这种情况下，只需稍微修改一下你的逻辑就可以创造奇迹。

TIMESTAMPDIFF(day,some_timestamp,@now) <= 45

可以改为

some_timestamp >= @now - INTERVAL 45 DAY

等等与其余的比较。通过将计算移出字段，可以使用索引。

<小时/>

此外，您也不需要多个查询和 UNION。

SELECT CASE WHEN some_timestamp >= @now - INTERVAL 45 DAY THEN "45 Days"
            WHEN some_timestamp >= @now - INTERVAL 90 DAY THEN "90 Days"
            ELSE "the rest"
       END AS `interval`
     , COUNT(*) AS `theCount`
FROM blah.yada
GROUP BY `interval`
;