Dimension with 200K+ members

[VivekD]

We are finalizing on a design for financial forecasting model in SAC and as part of that we have a dimension that has 200K+ members and that dimension is also used to control data access. 

This is basically our central organization dimension where we map all SAP WBSs, Cost Centers, SFDC Deals etc to our Account/Org structure. It will be used pretty extensively for filtering, reporting etc so want to make sure that SAC can handle such large volumes. 

Is having a large dimension like this a problem in terms of performance? 

[pedrosan · Accepted solution]

Hi @VivekD ,

SAC help defines dimension members limit as 1 million however you might encounter performance issues above 150k specially if you are using parent child hierarchies or geo enrichment.

I have already worked with dimensions without a hierarchy a bit over 1M members and we got to manage the performance properly however it is always recommended to try and keep the dimension members as low as possible specially if your performance expectations are too high(callculations to run below 15 secs sort of). Any chance you can split your model and therefore keep the master data volume lower?

I recommend checking the following links:

Managing Performance in SAP Analytics Cloud

https://help.sap.com/docs/SAP_ANALYTICS_CLOUD/00f68c2e08b941f081002fd3691d86a7/5f61b6509a5e4d499e8cb...

https://help.sap.com/docs/SAP_ANALYTICS_CLOUD/00f68c2e08b941f081002fd3691d86a7/25709793593a459d928ee...

Kind regards,

Pedro

[VivekD · Accepted solution]

Thanks for the detailed response. We do have parent-child hierarchies in the dimension. Performance expectations are mainly from story based consumption of the data from a user standpoint. We would ideally like all interactions to be within 1-2 secs.

Are level-based hierarchies better as opposed to parent-child hierarchies when it comes to performance?