Question

I have a fairly complex query (it takes about 30 seconds to execute) that returns me the following dataset:

 Month     Buy     Sell
2010/10     1        2
2010/11     1        3
2010/12     2        5

And here’s the query:

select month, avg(buy) [buy], avg(sell) [sell] from products group by month order by month

Now I want to add two Accumulated Columns and the expected resultset is the following:

 Month   Ac. Buy   Ac. Sell
2010/10     1          2
2010/11     2          5
2010/12     4          10

I’m trying to use this query

select 
distinct x.month
,(select SUM(buy) from products where month <= x.month) [Ac Buy]
,(select SUM(sell) from products where month <= x.month) [Ac Sell]
from products X
order by x.month

But that takes way too long!

Is there any way to do this faster?

I’m using a MS SQL 2008 Server, but my compability level is set to 80 (like MSSQL 2000, and I can’t change that). So I feel like I’m driving a Ferrari using only the 1st gear. );

Answer 1

For 13 rows I would just materialise the intermediate results into a table variable then do a triangular join on that.

DECLARE @Results TABLE
(
Month char(7) PRIMARY KEY,
Buy int,
Sell int
)

INSERT INTO @Results /*Your select query goes here*/
SELECT '2010/10',1,2 UNION ALL
SELECT '2010/11',1,3 UNION ALL
SELECT '2010/12',2,5

SELECT R1.Month,
       R1.Buy,
       R1.Sell,
       SUM (R2.Sell)AS AcSell,
       SUM (R2.Buy) AS AcBuy 
FROM @Results R1
JOIN @Results R2 ON R2.Month <= R1.Month
GROUP BY R1.Month,
       R1.Buy,
       R1.Sell