枢轴VS案例T-sql效率

我想弄清楚哪种方式更有效率。我试图使用Sql小提琴，但它似乎没有识别PIVOT或UNION 我可以查看哪些工具或统计数据来确定哪些更有效。枢轴VS案例T-sql效率

CREATE TABLE T 
(
    PersonNum INT 
,WeekOf DATETIME 
,ActivityType1 INT 
,ActivityType2 INT 
,Hours INT 
) 

INSERT INTO T VALUES(1,'11/2/2014',5,0,40) 
INSERT INTO T VALUES(1,'11/2/2014',1,0,5) 
INSERT INTO T VALUES(1,'11/2/2014',1,0,8) 
INSERT INTO T VALUES(1,'11/2/2014',2,1,6) 
INSERT INTO T VALUES(1,'11/2/2014',2,2,2) 
INSERT INTO T VALUES(1,'11/2/2014',2,3,9) 
INSERT INTO T VALUES(1,'11/2/2014',2,4,7) 

INSERT INTO T VALUES(1,'11/9/2014',5,0,40) 
INSERT INTO T VALUES(1,'11/9/2014',1,0,2) 
INSERT INTO T VALUES(1,'11/9/2014',1,0,6) 
INSERT INTO T VALUES(1,'11/9/2014',2,1,7) 
INSERT INTO T VALUES(1,'11/9/2014',2,2,2) 
INSERT INTO T VALUES(1,'11/9/2014',2,3,3) 
INSERT INTO T VALUES(1,'11/9/2014',2,4,5)

方法1

SELECT 
    PersonNum 
, WeekOf 
, SUM(CASE WHEN ActivityType1 = 5 THEN Hours ELSE 0 END) AS Beginning 
, SUM(CASE WHEN ActivityType1 = 1 THEN Hours ELSE 0 END) AS Plus 
, SUM(CASE WHEN ActivityType1 = 2 AND ActivityType2 <> 3 THEN Hours ELSE 0 END) AS Minus 
, SUM(CASE WHEN ActivityType1 = 2 AND ActivityType2 = 3 THEN Hours ELSE 0 END) AS MinusSpecial 
FROM T 
GROUP BY 
    PersonNum 
, WeekOf

方法2

SELECT 
    PersonNum 
, WeekOf 
, [Beginning] 
, [Plus] 
, [Minus] 
, [Minus Special] 
FROM 
(
    SELECT 
    PersonNum 
    , WeekOf 
    , 'Beginning' AS ColumnType 
    , Hours 
    FROM T 
    WHERE 
    ActivityType1 = 5 
    UNION 
    SELECT 
    PersonNum 
    , WeekOf 
    , 'Plus' AS ColumnType 
    , Hours 
    FROM T 
    WHERE 
    ActivityType1 = 1 
    UNION 
    SELECT 
    PersonNum 
    , WeekOf 
    , 'Minus' AS ColumnType 
    , Hours 
    FROM T 
    WHERE 
     ActivityType1 = 2 
    AND ActivityType2 <> 3 
    UNION 
    SELECT 
    PersonNum 
    , WeekOf 
    , 'Minus Special' AS ColumnType 
    , Hours 
    FROM T 
    WHERE 
     ActivityType1 = 2 
    AND ActivityType2 = 3 
    ) Data 
PIVOT 
(SUM(Hours) 
FOR ColumnType IN ([Beginning]     
        ,[Plus] 
        ,[Minus] 
        ,[Minus Special]) 
) pvt

来源

2014-11-14 Mike

看看这篇文章。它涵盖了执行此类查询的不同方式并探讨了性能。 http://www.sqlservercentral.com/articles/T-SQL/63681/ – 2014-11-14 19:29:17

你可以做到这一点

set statistics time on 

--first query 

set statistics time off 

set statistics time on 

--second query 

set statistics time off

你可以看一下在消息窗口中看到执行时间

来源

2014-11-14 18:47:01 CSharper

哇，这个数据的速度大约慢了3.5倍 – Mike 2014-11-14 19:09:01

方法1肯定会更快。这是因为只有一条select语句。另一方面，方法2中有5条select语句。一般规则是，查询中包含的语句越多，执行的时间就越长。

来源

2014-11-17 09:35:43 scar80

我可以建议重写你的PIVOT查询，所以你不必打表5次。

select P.PersonNum, 
     P.WeekOf, 
     P.Beginning, 
     P.Plus, 
     P.Minus, 
     P.MinusSpecial 
from (
    select T.PersonNum, 
      T.WeekOf, 
      T.Hours, 
      case ActivityType1 
      when 5 then 'Beginning' 
      when 1 then 'Plus' 
      when 2 then case ActivityType2 
          when 3 then 'MinusSpecial' 
          else 'Minus' 
         end 
      end as ColumnType 
    from T 
    ) as T 
pivot (sum(T.Hours) for T.ColumnType in (Beginning, Plus, MinusSpecial, Minus)) as P

因此，您可能会通过查询和大致相同的性能获得与您的组相同的查询计划。

来源

2014-11-17 09:49:31

枢轴VS案例T-sql效率

回答

相关问题