Size constraints of BW InfoProviders

[Former Member]

Hi,

While designing a new Data Model in BW 7.X- what is the size constraints of BW InfoProviders that needs to be considered ?

Suppose I have to bring historical data in BW of around 7 years, what is the size limitation in terms of -

a) Records in PSA table

b) Records in DSO

c) Records in InfoCube

d) Ratio of dimension table to fact table (in terms of Records count)

e) Query Throughput

f) Query Navigation Steps

Are there any standard statistics in terms of Memory allocation(GB) for the DSO, InfoCube etc or in total ?

Regards,

Amogh

[former_member181964]

Hi,

If you give all inputs in Quick Sizer in www.service.sap.com then it will prapose the complete hardware sizing etc..

So check it.

Thanks

Reddy

[Naga_P]

Hi Amogh,

I am not sure any specific benchmarks for each of these objects. But, generally we make sure each cube (the fact tables) would not be over 1 GB in size. We, in our current project, process around 2 to 3 millions at a go. If we have more than that we process them as two seperate loads. But, general rule is - the more the number of records in a PSA table, the less would be the loading performance.

The same applies to the ODS tables also - especially the Change Log table. Huge change log tables reduce the data loading performance.

About the ration of the Dimension table entries to the fact table is 1 : 10... meaning for every records in a dimension table, we should have at the least 10 records in the fact table. Any ratio less than this, reduces the performance of the objects - any where between 'negligible' to 'drastic' - depending the server resources, amount of data in the cube and so on.

rgds

Naga