DB2 Union causes time out? - sql

Question

First time working with DB2. I have a stored proc that performs a union, well it trys to anyhow. The procedure times out. The select statements on either side of the union execute quickly with no issue when done individually. Why would a union do this?
Here is some sudo code for the proc, assume data types are either int or string, and they match when compared or unioned unless otherwise noted in the SQL:
DECLARE Foo CURSOR WITH RETURN FOR
Select STRIP(A.Name) as my_Name,
Case A.Number
when 2 then '(' || strip(char(A.Number)) || ')' strip(B.num)
when 3 then '(' || strip(char(A.Number)) || ')' strip(C.num)
when 4 then '(' || strip(char(A.Number)) || ')' strip(D.num)
when 5 then '(' || strip(char(A.Number)) || ')' strip(E.num)
when 6 then '(' || strip(char(A.Number)) || ')' strip(F.num)
end as my_number
FROM A
left outer join B on A.Number= 2 and A.Name = B.Name
left outer join C on A.Number= 3 and C.Name = B.Name
left outer join D on A.Number= 4 and D.Name = B.Name
left outer join E on A.Number= 5 and E.Name = B.Name
left outer join F on A.Number= 6 and F.Name = B.Name
,session.Temp_Result X
WHERE X.ID = A.ID
GROUP BY A.Number, A.Name, B.Name, C.Name, D.Name, E.Name, F.Name
for fetch only ur;
DROP TABLE Session.Temp_Result;
DECLARE GLOBAL TEMPORARY TABLE session.Temp_Result
( ID DECIMAL(18,0)
);
INSERT INTO session.Temp_Result
select X.ID
from Z, Y, X, Q
where Z.num = 6
and Z.ID = Y.ID2
and Y.GROUPA = 'ABC'
and Y.GROUPB = 'DEF'
and Y.ID = X.ID2
and X.ID = Q.ID
union
select W.ID
from Z, Y, W
where Z.num = 6
and Z.ID = Y.ID2
and Y.GROUPA = 'ABC'
and Y.GROUPB = 'DEF'
and Y.ID = W.ID2
group by ID;
OPEN C_HIERARCHIES;

Answer 1

Try to query using "with" statement.
DECLARE GLOBAL TEMPORARY TABLE session.Temp_Result (
ID DECIMAL(18,0))
WITH REPLACE;
INSERT INTO session.Temp_Result
(ID)
WITH Q1(Y_ID)
AS (
SELECT Y.ID
FROM Z
INNER JOIN Y ON Z.ID = Y.ID2
WHERE Z.NUM = 6 AND Y.GROUPA = 'ABC' AND Y.GROUPB = 'DEF')
SELECT X.ID
FROM X
INNER JOIN Q1 ON X.ID2 = Q1.Y_ID
WHERE EXISTS(SELECT 1 FROM Q WHERE Q.ID = X.ID)
UNION
SELECT DISTINCT W.ID
FROM W
INNER JOIN Q1 ON W.ID2 = Q1.Y_ID
If it does not help, try to use "Explain SQL" for the select statement. Maybe you should create some indexes...

Answer 2

My guess is that the original code is missing a join condition that is present in the pseudocode. You should rewrite the query using proper join syntax to verify the query.
In any case, if each subquery runs quickly and there is some perverse consequence of a union, can you try splitting it into two inserts:
INSERT INTO session.Temp_Result
select X.ID
from Z, Y, X, Q
where Z.num = 6
and Z.ID = Y.ID2
and Y.GROUPA = 'ABC'
and Y.GROUPB = 'DEF'
and Y.ID = X.ID2
and X.ID = Q.ID;
INSERT INTO session.Temp_Result
select W.ID
from Z, Y, W
where Z.num = 6
and Z.ID = Y.ID2
and Y.GROUPA = 'ABC'
and Y.GROUPB = 'DEF'
and Y.ID = W.ID2
group by ID;
Then you can query Temp_Result as:
select distinct id from TempResult
Removing duplicates seems unnecessary, since the data is so small, but you could also do that directly.