Question

我有一个用于设备“heartbeats”的日志记录表。我有这些网络设备每10分钟与服务器签入/心跳。我们想要了解他们何时错过预定登记入住时间的统计数据。我有一个查询，可以在每个设备的基础上执行此操作，但我需要对其进行修改以处理所有设备。

心跳表如下所示：

CREATE TABLE [dbo].[DeviceHeartbeat](
    [Id] [int] IDENTITY(1,1) NOT NULL,
    [DeviceId] [int] NULL,
    [CheckinTime] [datetime] NULL,
    [Runtime] [int] NULL,
PRIMARY KEY CLUSTERED 
([Id] ASC)) ON [PRIMARY]

设备检查服务器，服务器向该表添加一行，包括其Id，CheckinTime和设备的运行时（设备发送的硬件值）。我目前的查询看起来像这样：

WITH t AS
(
  SELECT Checkintime, rn = ROW_NUMBER() OVER (ORDER BY Checkintime)
  FROM DeviceHeartbeat
  WHERE DeviceId = 1112
),
x AS
(
  SELECT d = DATEDIFF(MINUTE, t1.Checkintime, t2.Checkintime)
  FROM t AS t1
  INNER JOIN t AS t2
  ON t1.rn = t2.rn - 1
),
y AS
(
  SELECT stats = CASE WHEN d < 10 THEN ' < 10 '
    WHEN d BETWEEN 10 AND 11 THEN '10 - 11 '
    WHEN d BETWEEN 11 AND 12 THEN '11 - 12 '
    ELSE '+12 ' END + ' minutes:'
  FROM x
)
SELECT stats, COUNT(*) FROM y GROUP BY stats;

此查询仅限于单个指定的设备。示例结果如下所示：

stats                  
----------------- ---- 
 < 10  minutes:   1536
10 - 11  minutes: 425
11 - 12  minutes: 952
+12  minutes:     160

理想情况下，我只关心超过12分钟的签到。所以，我想要的是一个设备列表，这些设备的签到时间超过12分钟，按其计数排序。这将允许我查看超过12分钟登记时间的前10或20个设备，提醒我有问题的设备。类似的东西：

DeviceId   CheckinsOver12Mins
---------- -------------------
1112       160
1108       152
15         114
106        86

建议？

Answer 1

试试这个：

WITH t AS
(
  SELECT Checkintime, DeviceID, rn = ROW_NUMBER() OVER (ORDER BY DeviceID, Checkintime)
  FROM DeviceHeartbeat
),
x AS
(
  SELECT t1.deviceID, d = DATEDIFF(MINUTE, t1.Checkintime, t2.Checkintime)
  FROM t AS t1
  INNER JOIN t AS t2
  ON t1.rn = t2.rn - 1 and t1.DeviceID = t2.DeviceID
),
y AS
(
  SELECT deviceID
  FROM x
  WHERE d > 12
)
select deviceID, count(deviceID) as [Checkins over 12 mins] FROM y GROUP BY deviceID

注意：没有测试数据 - 没有测试，可能有拼写错误。

应该删除y CTE并将其更改为使其成为较小的查询：

select deviceID, count(deviceID) as [Checkins over 12 mins] 
FROM x 
GROUP BY deviceID
HAVING d > 12

SQL Query聚合DateDiff大于给定值

1 个答案: