Reputation: 63
I am using SQL Server 2012 Standard and have a performance relation to merge multiple rows into one.
Example:
I can use below query to get data as expected but the performance is not good. Is there any other query with better performance?
WITH Data AS
(
SELECT 88 ID, 1 AS [OrderFW],'a' AS Code,'n1' as Name UNION
SELECT 88 ID,2 AS [OrderFW],'a' AS Code,'n2' as Name UNION
SELECT 88 ID,3 AS [OrderFW],'a' AS Code,'n3' as Name UNION
SELECT 99 ID,1 AS [OrderFW],'b' AS Code,'n4' as Name UNION
SELECT 99 ID,2 AS [OrderFW],'b' AS Code,'n5' as Name
)
SELECT
d1.Code code1, d1.Name name1,
d2.Code code2, d2.Name name2,
d3.Code code3, d3.Name name3
FROM
Data d1
LEFT OUTER JOIN
Data d2 ON d1.ID = d2.ID AND d2.OrderFW = 2
LEFT OUTER JOIN
Data d3 ON d1.ID = d3.ID AND d3.OrderFW = 3
WHERE
d1.OrderFW = 1
Upvotes: 1
Views: 1087
Reputation: 50173
I would try with aggregation:
select Id,
max(case when seq = 1 then code end) as code1,
max(case when seq = 1 then name end) as name1,
max(case when seq = 2 then code end) as code2,
max(case when seq = 2 then name end) as name2,
max(case when seq = 3 then code end) as code3,
max(case when seq = 3 then name end) as name3
from (select d.*,
row_number() over (partition by code order by order) as seq
from data d
) d
group by Id;
EDIT : If you have a already sequence
(i.e. Order
) then only aggregation is enough :
select Id,
max(case when [order] = 1 then code end) as code1,
max(case when [order] = 1 then name end) as name1,
max(case when [order] = 2 then code end) as code2,
max(case when [order] = 2 then name end) as name2,
max(case when [order] = 3 then code end) as code3,
max(case when [order] = 3 then name end) as name3
from data d
group by Id;
EDIT : Thanks to JohnLBevan for Demo.
Here is SQL Fiddle Example.
Upvotes: 3