Skip to Content
0
Former Member
Apr 23, 2009 at 04:40 PM

Bad Performance Caused By Characteristic

45 Views

We have an area that its suffering of serious problem of performance, an example it is a cube that has 6M records and it barely output reports of 1 sales office (10% of the total in a minute or so) and its has an agreggated by the Client dimension taht we think maybe the problem.

We have a high performance machine, not problem there, in fact in another information area over SAP FI-AR, theres a cube (customer created) with 9M of records and it can have a full report without slice in a couple of seconds.

So back to the example cube, analizing its cardinality the Client (ZCLIENT) has about 20% of the fact table, explain by the total of clients that are about 1,7M, still we use high cardinality when this IO is present. This Client its not present in the areas that runs ok, there we use standard IO (0BPARTNER, 0CUSTOMER, etc.) with no more of 250.000 Records each.

We think that the problems come by the fact that this ZCLIENT that its present in all the cube with performance problems has about 50 atributes, and an average of 10 ATTR acitvated by cube, besides the fact of the 1,7M of records.

Can anybody give some modelling tips about this subjects?

any tip will be apreciated.