Subquery with TOP performing poorly

Question

Ok I have a performance problem in SQL function. I have query like this:

SELECT
    Number,
    (SELECT TOP 1 Value 
     FROM Table_B 
     WHERE Table_B.Number = Table_A.Number 
     ORDER BY ID DESC)
FROM 
    Table_A

I want see all records from Table_A with the TOP 1 record from Table_B

Answer 1

You may try adding the following index to Table_B :

CREATE INDEX idx_b ON Table_B (Number, ID) INCLUDE (Value);

This index should let the Number lookup be fast, and should allow the first Value , as sorted by the ID , to be found. We include Value at the end of the index since it is used in the select, but not needed in the index itself.

Answer 2

Use ROW_NUMBER() for serializing value with id desc and then LEFT JOIN main table with matching number and retrieve only first record from subquery result. LEFT JOIN is used here because table_B have no value for a number but as per requirement corresponding value of table_A must come.

-- SQL Server
SELECT a.Number
     , b.value
FROM Table_A a
LEFT JOIN (SELECT Number, ID, value
                , ROW_NUMBER() OVER (PARTITION BY Number, ID ORDER BY id DESC) row_num
           FROM Table_B) b         
       ON a.Number = b.Number
      AND b.row_num = 1

Answer 3

This is the "first in group" method that SMor mentions:

WITH cte AS
(
   SELECT a.Number,
          b.Value, 
          ROW_NUMBER() OVER (PARTITION BY a.Number ORDER BY b.ID DESC) AS rn
   FROM Table_A a
   LEFT OUTER JOIN Table_B b ON a.Number = b.Number
)
SELECT Number, Value
FROM cte
WHERE rn = 1 

Answer 4

WITH CTE_Table_B
AS(
    SELECT
        x.[ID],
        x.[Value],
        x.[Number]
    FROM(
        SELECT
            [ID]
            [Value],
            [Number],
            DENSE_RANK() OVER(PARTITION BY [Number] ORDER BY [ID] DESC) AS [Ranking]
        FROM Table_B
    ) AS x
    WHERE (1=1)
        AND x.[Ranking] = 1
)

SELECT
    a.[Number],
    b.[Value] 
FROM Table_A AS a

LEFT JOIN CTE_Table_B AS b
    ON b.[Number] = a.[Number]

Answer 5

How about just using CROSS APPLY (assuming there will always be a match - if not use OUTER APPLY):

SELECT
    Number, tb.Value
FROM 
    Table_A
CROSS APPLY
    (SELECT TOP 1 Value 
     FROM Table_B 
     WHERE Table_B.Number = Table_A.Number 
     ORDER BY ID DESC) AS tb