Sql 子查询中的CTE

我有一个场景显示来自两个表的数据。第一个名为“DayTable”的表由每日计划和实际值组成。第二个名为“MONTTABLE”的表格由月度计划和实际数据组成。我需要显示最近6个月的数据和本月的每日数据。因此，我为预期的输出编写了如下所示的查询

 Declare @startdate date = CONVERT(DATE, DATEADD(dd, -DAY(DATEADD(MONTH, 0, GETDATE())) + 1, DATEADD(MONTH, 0, GETDATE()))) 
 Declare @endDate date = DATEADD(DAY, -DAY(DATEADD(MONTH, 1, GETDATE())), DATEADD(MONTH, 1, GETDATE()))

CREATE TABLE #TEMP
(
 PlanDate NVARCHAR(100),
 [PastTrend - Plan] INT,
 [PastTrend - Actual] INT,
 [Current - Plan] INT,
 [Current - Actual] INT,
)

;With cte
as
(
    Select @startdate sDate
    Union All
    Select DATEADD(day,1,sDate) From cte where DATEADD(day,1,sDate) <= @endDate
)
INSERT INTO #TEMP
SELECT 
REPLACE(CONVERT(CHAR(6), A.sDate, 106),' ',' - ') PlanDate 
,NULL AS [PastTrend - Plan]
,NULL AS [PastTrend - Actual]
,SUM(B.PlanQuantity) AS [Current - Plan]
,SUM(B.Actual) AS [Current - Actual]
FROM cte A 
LEFT OUTER JOIN DayTable B 
ON A.sDate = CONVERT(DATE,B.PlanDate) 
GROUP BY A.sDate 
--ORDER BY A.sDate


SELECT
*

FROM
(
SELECT 
CONVERT(CHAR(3), datename(month,PlanMonth)) + ' ' + RIGHT(CONVERT(VARCHAR(4), YEAR(PlanMonth)), 2)  AS PlanDate
,SUM(PlanQuantity) AS [PastTrend - Plan]
,SUM(Actual) AS [PastTrend - Actual]
,NULL AS [Current - Plan]
,NULL AS [Current - Actual]
FROM 
MonthTable 
WHERE CONVERT(DATE, PlanMonth) >= CONVERT(DATE, DATEADD(dd, -DAY(DATEADD(MONTH, 0, GETDATE())) + 1, DATEADD(MONTH, -6, GETDATE())))
group by PlanMonth 


UNION ALL

SELECT 
PlanDate
,[PastTrend - Plan]
,[PastTrend - Actual]
,[Current - Plan]
,[Current - Actual]
FROM 
#TEMP

) T1


 DROP TABLE #TEMP

我的输出是

现在我想避免临时表的概念，因为如果在创建临时表之后出现任何故障，它将不会下降。因此，像下面这样重写查询

 Declare @startdate date = CONVERT(DATE, DATEADD(dd, -DAY(DATEADD(MONTH, 0, GETDATE())) + 1, DATEADD(MONTH, 0, GETDATE()))) 
 Declare @endDate date = DATEADD(DAY, -DAY(DATEADD(MONTH, 1, GETDATE())), DATEADD(MONTH, 1, GETDATE()))

;With cte
as
(
    Select @startdate sDate
    Union All
    Select DATEADD(day,1,sDate) From cte where DATEADD(day,1,sDate) <= @endDate
)
SELECT 
A.sDate AS OriginalDate
,REPLACE(CONVERT(CHAR(6), A.sDate, 106),' ',' - ') PlanDate 
,NULL AS [PastTrend - Plan]
,NULL AS [PastTrend - Actual]
,SUM(B.PlanQuantity) AS [Current - Plan]
,SUM(B.Actual) AS [Current - Actual]
FROM cte A 
LEFT OUTER JOIN DayTable B 
ON A.sDate = CONVERT(DATE,B.PlanDate) 
GROUP BY A.sDate 

UNION ALL

SELECT 
PlanMonth AS OriginalDate
,CONVERT(CHAR(3), datename(month,PlanMonth)) + ' ' + RIGHT(CONVERT(VARCHAR(4), YEAR(PlanMonth)), 2)  AS PlanDate
,SUM(PlanQuantity) AS [PastTrend - Plan]
,SUM(Actual) AS [PastTrend - Actual]
,NULL AS [Current - Plan]
,NULL AS [Current - Actual]
FROM 
MonthTable 
WHERE CONVERT(DATE, PlanMonth) >= CONVERT(DATE, DATEADD(dd, -DAY(DATEADD(MONTH, 0, GETDATE())) + 1, DATEADD(MONTH, -6, GETDATE())))
group by PlanMonth 

ORDER BY OriginalDate

---

但我有个问题。在输出中，我不需要原始数据。如何避免这种情况。为此，我可以将union输出包装到select查询中，但如何在cte中获得错误。请引导我。还建议哪种方法是最好的

我完成了查询。实际上，我只是将查询包装为outer，不包含cte。Cte必须位于查询的顶部。最后一个查询是

    Declare @startdate date = CONVERT(DATE, DATEADD(dd, -DAY(DATEADD(MONTH, 0, GETDATE())) + 1, DATEADD(MONTH, 0, GETDATE()))) 
     Declare @endDate date = DATEADD(DAY, -DAY(DATEADD(MONTH, 1, GETDATE())), DATEADD(MONTH, 1, GETDATE()))



;With cte
    as
    (
        Select @startdate sDate
        Union All
        Select DATEADD(day,1,sDate) From cte where DATEADD(day,1,sDate) <= @endDate
    )
    SELECT
    T1.PlanDate
    ,T1.[PastTrend - Plan]
    ,T1.[PastTrend - Actual]
    ,T1.[Current - Plan]
    ,T1.[Current - Actual]
    FROM
    (
    SELECT 
    A.sDate AS OriginalDate
    ,REPLACE(CONVERT(CHAR(6), A.sDate, 106),' ',' - ') PlanDate 
    ,NULL AS [PastTrend - Plan]
    ,NULL AS [PastTrend - Actual]
    ,SUM(B.PlanQuantity) AS [Current - Plan]
    ,SUM(B.Actual) AS [Current - Actual]
    FROM cte A 
    LEFT OUTER JOIN DayTable B 
    ON A.sDate = CONVERT(DATE,B.PlanDate) 
    GROUP BY A.sDate 

    UNION ALL

    SELECT 
    PlanMonth AS OriginalDate
    ,CONVERT(CHAR(3), datename(month,PlanMonth)) + ' ' + RIGHT(CONVERT(VARCHAR(4), YEAR(PlanMonth)), 2)  AS PlanDate
    ,SUM(PlanQuantity) AS [PastTrend - Plan]
    ,SUM(Actual) AS [PastTrend - Actual]
    ,NULL AS [Current - Plan]
    ,NULL AS [Current - Actual]
    FROM 
    MonthTable 
    WHERE CONVERT(DATE, PlanMonth) >= CONVERT(DATE, DATEADD(dd, -DAY(DATEADD(MONTH, 0, GETDATE())) + 1, DATEADD(MONTH, -6, GETDATE())))
    group by PlanMonth 
    ) T1 
    ORDER BY T1.OriginalDate

但我需要知道性能。当我以相对于批次的实际执行计划查询成本执行此查询时：100%

---

当我使用temp执行第一个方法时，查询成本是90%。任何人都可以使用此guid吗

如果存在，为什么不将其包括在内。。。。去临时桌吗？您也可以考虑一个Test.catch逻辑来删除临时表，以防出现故障。但我正在努力避免临时表。我不知道哪种方法好。我认为直接法简单又好。同样像这样，我还有其他程序也使用temp。所以我想避免它。就这些。如果临时表是更好的方法，我可以这样做