Should db2 summarize or home grow a build app in C

[mmiikkee]

I have a large (perhaps small) table (foo) of 48 Million rows and I
have a temp table which I load with keys (2 to 3 Mil). I use the temp
table to extract and aggregate some columns from foo (basically by a
transaction date) that I then insert into a summary table. Now foo
might hold 10 rows to aggregate but the majority of the time (90%) the
sum extracts a single row. Thus the inserts usually match the temp row
count. The temp row length is 110 bytes but the keys are < 20 bytes.
Foo rows are 310 bytes. This is a udb2 on 40 processors. I'm sure
much larger summations are performed in business. Is it unreasonable
to have the engine do this type of work? The reason is I seem to be
getting an sql955. My feeling is this should not be a big deal.
Thanks for any comments.

[Ian]

mm******@att.net wrote:

I have a large (perhaps small) table (foo) of 48 Million rows and I
have a temp table which I load with keys (2 to 3 Mil). I use the temp
table to extract and aggregate some columns from foo (basically by a
transaction date) that I then insert into a summary table. Now foo
might hold 10 rows to aggregate but the majority of the time (90%) the
sum extracts a single row. Thus the inserts usually match the temp row
count. The temp row length is 110 bytes but the keys are < 20 bytes.
Foo rows are 310 bytes. This is a udb2 on 40 processors. I'm sure
much larger summations are performed in business. Is it unreasonable
to have the engine do this type of work? The reason is I seem to be
getting an sql955. My feeling is this should not be a big deal.
Thanks for any comments.

This is not an unreasonable task for DB2 to perform!

However, you should work with your DBAs to configure the sort
memory allocations properly. If you're getting SQL0955N errors,
the memory allocation has probably not been configured properly
for your environment / application.

[mmiikkee]

Thanks Ian, I'll take your advice.