我的数据库中有以下属性。
统计ID,设备名称,值,时间戳。
对于给定的统计信息,我想找到每个唯一设备的2个最新时间戳和相应的值。
我正在尝试像
这样的东西试验1)
select statistic, devicename, value, timestamp
from X_STATSVALUE
where statistic=19
order by orgtime DESC limit 2;
这为我提供了前2个时间戳,但不是每个设备。
试验2)
select statistic, devicename, value, timestamp
from X_STATSVALUE as x
where x.statistic=241
and (select count(*)
from X_STATSVALUE as y
where y.statistic=x.statistic
and y.device=x.device
and y.timestamp > x.timestamp) <=1;
但这也不太好..
基本上,对于给定的统计信息,我希望最新的2个时间戳记包含每个设备的值..任何帮助都非常感谢:)
答案 0 :(得分:2)
这就是我解决这类问题的方法:
SELECT x.statistic, x.devicename, x.value, x.timestamp
FROM X_STATSVALUE AS x
LEFT OUTER JOIN X_STATSVALUE AS x2
ON (x.statistic = x2.statistic
AND x.devicename = x2.devicename
AND x.timestamp < x2.timestamp)
GROUP BY x.statistic, x.devicename
HAVING COUNT(*) < 2;
换句话说,显示行,使得少于两个具有相同statistic
和devicename
以及更大(更近期)timestamp
的行。
我假设您在timestamp
列中没有给定统计数据和副本的重复项设备名。
答案 1 :(得分:1)
尝试类似:
SELECT DISTINCT x.statistic, x.devicename, x.value, x.timestamp
FROM X_STATSVALUE AS x
WHERE x.timestamp IN (SELECT timestamp
FROM X_STATSVALUE
WHERE devicename = x.devicename AND statistic = x.statistic
ORDER BY timestamp LIMIT 2)
(虽然可能不适用于旧的MySQL)
答案 2 :(得分:1)
我在我的系统上测试了以下查询,并且像你的那样有一个演示表,它可以工作。我考虑过orgtime而不是时间戳...你可以做同样的事情(只需更改姓名)
select t1.statistic, devicename, value, timestamp from X_STATSVALUE as t1
inner join
( select statistic, max(orgtime) as orgtime from X_STATSVALUE group by statistic )
as t2
on t1.statistic = t2.statistic and t1.orgtime = t2.orgtime
UNION
select tb1.statistic, tb1.devicename, tb1.value, tb1.timestamp
from X_STATSVALUE as tb1 inner join
( select statistic, max(orgtime) as orgtime from X_STATSVALUE WHERE statistic+orgtime not in
(select t1.statistic+t1.orgtime from X_STATSVALUE as t1 inner join
( select statistic, max(orgtime) as orgtime from X_STATSVALUE group by statistic )
as t2 on t1.statistic = t2.statistic and t1.orgtime = t2.orgtime
) group by statistic
)
as tb2 on tb1.statistic = tb2.statistic and tb1.orgtime = tb2.orgtime