有效地为集合中的每个类别选择顶行

Question

我需要从已知集合中为每个类别选择一个顶行（有点类似于this question）。问题是，如何让这个查询在大量行上高效。

例如，让我们创建一个表来存储多个位置的温度记录。

CREATE TABLE #t (
    placeId int,
    ts datetime,
    temp int,
    PRIMARY KEY (ts, placeId)
)

-- insert some sample data

SET NOCOUNT ON

DECLARE @n int, @ts datetime
SELECT @n = 1000, @ts = '2000-01-01'

WHILE (@n>0) BEGIN
    INSERT INTO #t VALUES (@n % 10, @ts, @n % 37)
    IF (@n % 10 = 0) SET @ts = DATEADD(hour, 1, @ts)
    SET @n = @n - 1
END

现在我需要获取位置 1、2、3 的最新记录。

这种方式很有效，但不能很好地扩展（而且看起来很脏）。

SELECT * FROM (
    SELECT TOP 1 placeId, temp
    FROM #t 
    WHERE placeId = 1
    ORDER BY ts DESC
) t1
UNION ALL
SELECT * FROM (
    SELECT TOP 1 placeId, temp
    FROM #t 
    WHERE placeId = 2
    ORDER BY ts DESC
) t2
UNION ALL
SELECT * FROM (
    SELECT TOP 1 placeId, temp
    FROM #t 
    WHERE placeId = 3
    ORDER BY ts DESC
) t3

以下看起来更好，但工作效率要低得多（根据优化器，30% 对 70%）。

SELECT placeId, ts, temp FROM (
    SELECT placeId, ts, temp, ROW_NUMBER() OVER (PARTITION BY placeId ORDER BY ts DESC) rownum
    FROM #t
    WHERE placeId IN (1, 2, 3)
) t
WHERE rownum = 1

问题是，在后一个查询执行计划中，对 #t 执行聚集索引扫描，检索、排序、编号和过滤 300 行，只剩下 3 行。对于前一个查询，3 次获取一行。

有没有办法在没有大量联合的情况下高效地执行查询？

Answer 1

不要只看执行计划还要看statistics io和statistics time

set statistics io on
go
SELECT * FROM (
    SELECT TOP 1 placeId, temp
    FROM #t 
    WHERE placeId = 1
    ORDER BY ts DESC
) t1
UNION ALL
SELECT * FROM (
    SELECT TOP 1 placeId, temp
    FROM #t 
    WHERE placeId = 2
    ORDER BY ts DESC
) t2
UNION ALL
SELECT * FROM (
    SELECT TOP 1 placeId, temp
    FROM #t 
    WHERE placeId = 3
    ORDER BY ts DESC
) t3

SELECT placeId,  temp FROM (
    SELECT placeId, ts, temp, ROW_NUMBER() OVER (PARTITION BY placeId ORDER BY ts DESC) rownum
    FROM #t
    WHERE placeId IN (1, 2, 3)
) t
WHERE rownum = 1

set statistics io off
go

表“#t000000000B99”。扫描计数 3，逻辑读取 6，物理读取 0，预读读取 0，lob 逻辑读取 0，lob 物理读取 0，lob 预读读取 0。 表'#t000000000B99'。扫描计数 1，逻辑读取 6，物理读取 0，预读读取 0，lob 逻辑读取 0，lob 物理读取 0，lob 预读读取 0。

set statistics time on
go
SELECT * FROM (
    SELECT TOP 1 placeId, temp
    FROM #t 
    WHERE placeId = 1
    ORDER BY ts DESC
) t1
UNION ALL
SELECT * FROM (
    SELECT TOP 1 placeId, temp
    FROM #t 
    WHERE placeId = 2
    ORDER BY ts DESC
) t2
UNION ALL
SELECT * FROM (
    SELECT TOP 1 placeId, temp
    FROM #t 
    WHERE placeId = 3
    ORDER BY ts DESC
) t3

SELECT placeId,  temp FROM (
    SELECT placeId, ts, temp, ROW_NUMBER() OVER (PARTITION BY placeId ORDER BY ts DESC) rownum
    FROM #t
    WHERE placeId IN (1, 2, 3)
) t
WHERE rownum = 1

set statistics time on
go

对我来说这两种方法没有真正的区别，加载更多数据并再次比较

此外，当您向两个查询添加 order by 时，它会下降到 40% 和 60%

SELECT * FROM (
    SELECT TOP 1 placeId, temp
    FROM #t 
    WHERE placeId = 1
    ORDER BY ts DESC
) t1
UNION ALL
SELECT * FROM (
    SELECT TOP 1 placeId, temp
    FROM #t 
    WHERE placeId = 2
    ORDER BY ts DESC
) t2
UNION ALL
SELECT * FROM (
    SELECT TOP 1 placeId, temp
    FROM #t 
    WHERE placeId = 3
    ORDER BY ts DESC
) t3
ORDER BY placeId

SELECT placeId,  temp FROM (
    SELECT placeId,  temp, ROW_NUMBER() OVER (PARTITION BY placeId ORDER BY ts DESC) rownum
    FROM #t
    WHERE placeId IN (1, 2, 3)
) t
WHERE rownum = 1
ORDER BY placeId

Answer 2

我加载了 100,000 行（这仍然不足以减慢速度），尝试了老式的方法：

select t.*
 from #t t
  inner join (select placeId, max(ts) ts
               from #t
               where placeId in (1,2,3)
               group by placeId) xx
   on xx.placeId = t.placeId
    and xx.ts = t.ts

得到了几乎相同的结果。

然后我将索引中列的顺序颠倒，以

CREATE TABLE #t ( 
    placeId int, 
    ts datetime, 
    temp int, 
    PRIMARY KEY (placeId, ts) 
) 

并且，在所有查询中，页面读取和索引搜索而不是扫描。

如果优化是您的目标并且您可以修改索引，我会修改主键，或者添加一个覆盖索引。

Answer 3

仅作记录，使用 CROSS APPLY 的另一个选项。
在我的配置中，它的性能比之前提到的所有配置都要好。

SELECT *
FROM (VALUES (1),(2),(3)) t (placeId)
CROSS APPLY (
    SELECT TOP 1 ts, temp
    FROM #t 
    WHERE placeId = t.placeId
    ORDER BY ts DESC
) tt

我猜，VALUES 可以更改为临时表或表变量，没有太大区别。