Question:
I am currently writing a program that calculates the size of any or all cubes in our systems. I just require some confirmation and/or suggestions to make sure my logic is correct and hopefully I won't start some heated debate over cube architecture.
The cubes size is essentially the sizes of the fact table (/BI#/F####) and the dimension table (/BI#/D####). Now should I exclude the other tables for infoobjects: SID, Hier, Text, M Table or include them in the size. I personally think I should exclude them and report on them separately as they are cube independant tables.
Am I missing any cube tables that any knows of? I know there's the E fact table but I don't believe that is populated after the data is activated.
Regards,
EL
Answer:
U can check the file size of the cube using table name called RSRCUBEDATALN and RSRCUBEDATA
HOPE THIS HELPS
Answer:
Dear Guest,
I'm not sure what those tables are for but they point to .csv files related to certain cubes. I cannot find any cubes that we have created ourselves. Perhaps this is for flat file loads?
Regards,
EL
Answer:
Did u check the table RSDCUBE???
Answer:
The E Fact table only contains data if you compress the cube and the database catalogue (DB2 EEE) or the oracle v$ tables will provide information on the size of the tables
Answer:
hi EL,
does your program work? If so, maybe you can give me a hint how it works...
I also got to determine the size of our cubes and I'm looking for the fastest, accurate enough way to do that.
Thanks,
Alex
Answer:
There is an SAP spreadsheet that calculates sizing based on entry of all the dimensions and characteristics. I think it is part of ASAP documentation; maybe this would provide the information you need, or would help you test your program. Depending on your aim, you may need to take account of indexes.
Answer:
EL,
Why not use transaction DB02 for this? With the detailed analysis you can see the actual size by selecting tables /BIC/*{cubename}*
If you want to have it all, just don't specify the tablespace name. The search will take some time then while BW will grasp all tablespaces. For a non-accurate but fair count, you can select tablespace PSAPFACTD for data and PSAPFACTI for the indexes.
Answer:
For ODS objects it is PSAPODSI and PSAPODSD I believe.