6mb Mdc using 80G Temp file to Create??

[FIR1987]

Hi

I have created a cube about 6mb, however, it generate about 80G Temp file. Could you tell what make the temporary file so big?

My data source is IQD, would it be smaller if I use cognos package or cognos report as my data source?

Anybody have any suggestions?

Thanks

[cognostechie]

Are there joins in the IQD ?

[redmist]

I have seen this happening if you have enabled Local processing. Is there any calculation that require local processing?

[cognostechie]

Correct ! That's why I asked if there are any joins in the IQD because it's possible that not all of the query can be processed in the DB depending on how the SQL is constructed.

[FIR1987]

Thanks , you  two really help me  a lot .If  so , I  think I could  replace local with server ,but i don 't know which calculation measure can't   be processed in the DB.

I have measure a in a.iqd and measure b in b.iqd ,and then I add a caculation measure using a+b .

Does it mean it  can't be processed in DB?

[cognostechie]

If you have two seperate IQDs and a calculated column in Transformer referencing both of them, that calculation can only be performed locally, not in the DB because two seperate SQLs are being sent to the DB, one for each IQD without any relation to them. When you build the cube, it sends the SQL of the IQD sequentially to the DB meaning that the 1st IQD will be sent to the DB and the 2nd one will go only after the 1st one gets done.

I don't think there is any way to stop that local processing unless you want to modify your IQDs which I would not recommend. However, that should still not grow the temp file to that limit. Do you have anything else that can cause this?