sql - 减少广泛的查询

Question

我正在从我们的数据库中收集工作单的历史数据。在过去的 12 个月中，我使用的是该月的最后一天。在那个日期，我想看看有多少票是开放的(datecreated 之前或等于一个月的最后一天和最后一天之后的 dateclose 或 null)。截至当月最后一天，这些工单中有多少开放时间超过 30 天，以及该月关闭了多少工单。

我的查询返回前 12 个月中每个月的 4 个值。我开发了一个查询，一次收集一个月的数据，然后使用 Union 获取下一个月的数据，依此类推。

我现在正在尝试开发一个查询，它不需要每个月收集的数据的联合。这是一个示例月。

Select Convert(Varchar,
               DATEADD(day,
                       -1,
                       DATEADD(mm, DATEDIFF(m, 0, GETDATE()) - 12, 0)),
               101) as [ DateRun ],
       t1. [ data1 ],
       sum(t2. [ data2 ] + t3. [ data3 ]) as [ Total Open ],
       (t1. [ data1 ] + 0.0) /
       sum((t2. [ data2 ] + 0.0) + (t3. [ data3 ] + 0.0)) as [ Percent Aged Over 30 Days ]
  From (Select count(*) as [ data1 ]
          From (Select s.ticketnumber, s.datecreated, s.dateclosed
                  from mydb2 i
                 inner join mydb1 s
                    on i.ticketNumber = s.ticketNumber
                 Where datediff(dd,
                                s.dateCreated,
                                DATEADD(day,
                                        -1,
                                        DATEADD(mm,
                                                DATEDIFF(m, 0, GETDATE()) - 12,
                                                0))) > '30'
                   AND (s.dateclosed is null OR
                        s.dateclosed >=
                        DATEADD(day,
                                -1,
                                DATEADD(mm, DATEDIFF(m, 0, GETDATE()) - 12, 0)))
                 Group by s.ticketNumber, s.datecreated, s.dateclosed) as tb1) as t1,
       (Select count(*) as [ data2 ]
          From (Select s.ticketnumber, s.datecreated, s.dateclosed
                  from mydb2 i
                 inner join mydb1 s
                    on i.ticketNumber = s.ticketNumber
                 Where s.datecreated <
                       DATEADD(day,
                               -1,
                               DATEADD(mm, DATEDIFF(m, 0, GETDATE()) - 12, 0))
                   AND (s.dateclosed is null OR
                       s.dateclosed >=
                       DATEADD(day,
                                -1,
                                DATEADD(mm, DATEDIFF(m, 0, GETDATE()) - 12, 0)))
                 Group by s.ticketNumber, s.datecreated, s.dateclosed) as tb2) as t2,
       (Select count(*) as [ data3 ]
          From (Select s.ticketnumber, s.datecreated, s.dateclosed
                  from mydb2 i
                 inner join mydb1 s
                    on i.ticketNumber = s.ticketNumber
                 Where s.dateclosed >
                       DATEADD(day,
                               -1,
                               DATEADD(mm, DATEDIFF(m, 0, GETDATE()) - 13, 1))
                   AND s.dateclosed <
                       DATEADD(day,
                               -1,
                               DATEADD(mm, DATEDIFF(m, 0, GETDATE()) - 12, 0))
                 Group by s.ticketNumber, s.datecreated, s.dateclosed) as tb3) as t3
 Group by t1. [ data1 ]

通过研究，我能够在最后一天使用以下代码获得过去 12 个月的数据，但我无法更新上面的代码，因此我不需要 12 个联合。

Select dateadd(month, 1 + datediff(month, 0, s.datecreated), -1) as [ Date Run ]
  from mydb1 s
 Where s.datecreated >
       DATEADD(day, -1, DATEADD(mm, DATEDIFF(m, 0, GETDATE()) - 12, 0))
   and s.datecreated <
       DATEADD(day, -1, DATEADD(mm, DATEDIFF(m, 0, GETDATE()), 0))
 Group by dateadd(month, 1 + datediff(month, 0, s.datecreated), -1)
 Order by 1

Answer 1

您可以尝试使用 CTE 生成月份，将其用作查询的基础，然后将聚合作为子查询运行。换句话说，像这样:

with Months( ClosingDate, n ) as
(
    select DATEADD(dd, -1, DATEADD(MONTH, DATEDIFF(MONTH, 0, GETDATE()), 0)) as ClosingDate, 1 as n
    union all
    select DATEADD(dd, -1, DATEADD(mm, -1, DATEADD(dd, 1, ClosingDate))), n + 1
    from Months
    where n < 12
) 
select DATENAME(mm, m.ClosingDate) MonthLabel, DATEADD(dd, 1 - DATEPART(dd, m.ClosingDate), m.ClosingDate) OpeningDate, m.ClosingDate,
ISNULL((select COUNT(*)
    from mydb1 s
    where s.dateCreated < DATEADD(dd, -29, m.ClosingDate)
      and (s.dateClosed is null or s.dateClosed >= DATEADD(dd, 1, m.ClosingDate))
), 0) [Total Open 30+ Days]
from Months m

为清楚起见，我仅包含其中一个聚合，并省略了您的连接和分组。