Question

I have a customer table and an orderdetail table
the customer id is a foriegn key in the orderdetail table (I’m simplifying here)

The orderdetail table contains the following columns
OrderId
ItemId
CustomerId
Size

The Size column can take on any one of the following values
1. lr
2. md
3. sm

So the orderdetail table could have the following records (I’ve comma delimited the columns)

OrderId ItemId CustomerId Size

1,1,30,lr
1,1,30,md
1,1,30,sm
2,1,30,lr
2,1,30,md
3,1,30,lr
3,1,30,sm
4,1,30,lr
5,1,30,md
6,2,30,sm
7,3,30,md
8,3,30,lr

What I’d like is a really efficient query (there’s millions of records in the order details tables) that has the following output for a given customerId (30 in this case)

ItemId  SizeLr  SizeMd  SizeSm
1         4       3        2
2         0       0        1
3         1       1        0

The query I’m using uses 3 group by queries (one each for “lr”, “md” and “sm”) and as a result it scans the table 3 times.

I’m looking for a solution that scans the table just onces. I think the solution is using the new Grouping Set feature in MSSQL 2008. But either ways, a solution using one scan if what I hope someone can help me with.

EDIT

The actual output needs to have other fields from the customer table and the order details table as well. These other fields are not dependent on the aggregates.

For example

ItemName  ItemId  SizeLr  SizeMd  SizeSm
   A         1      4       3       2
   B         2      0       0       1
   C         3      1       1       0
   Totals           5       4       4

It would nice if I could get the totals for each of the Size columns as well

Answer 1

How about this:

SELECT ItemId, 
    SUM(CASE WHEN Size = 'Lr' THEN 1 ELSE 0 END) AS SizeLr,
    SUM(CASE WHEN Size = 'Md' THEN 1 ELSE 0 END) AS SizeMd,
    SUM(CASE WHEN Size = 'Sm' THEN 1 ELSE 0 END) AS SizeSm
FROM OrderDetail 
WHERE CustomerId = 30
GROUP BY ItemId