cancel
Showing results for 
Search instead for 
Did you mean: 

Time dependent Objects

Former Member
0 Kudos

Hi,

We are trying to enable time dependent characteristics for some of the attributes for 0CUSTOMER. We are looking to know whether it will influence on performance going farward. I appreciate if you guys can suggest me other methods to get the reports( ex: creating on ODS object and for the required attributes and loading the changes to customer from R/3(OV51). I know getting extraction from tables related to OV51 transactioncode take longer time but our customer master database is very small from numbering point of view).

I will appreciate your early response.

Thanks in Advance

Radha Chalasani

Accepted Solutions (0)

Answers (2)

Answers (2)

Masaaki
Product and Topic Expert
Product and Topic Expert
0 Kudos

Hi, Radha,

In my experience, time depend attrebute influence report and staging peformance, especially when the number of that master data is large. And the master data that should have time depend attrebute tend to be important and have many record(more than 100,000 records). For example, customer, material, organization unit...

time depend attrebute influence the following issue.

- Report performance

- Staging performance(read that attrebute in process of staging)

- generating aggregation(aggregation that is aggregated by the time depend attrebute)

I have the experience that at first I have set attrebute as time depend and later found it influence the performance and added the same attrebute as time independent.(in this case, report is always executed as "today" and from report requirement, time depend attrebute is not required.)

Hope it helpes.

Kind regards,

Masaaki

Former Member
0 Kudos

One rule of thumb for performance is to utilize time dependent objects for "must have" requirements, not "nice to have", when possible. The reason is that a much larger number of values are often accumulated, thus making a larger number that need to be evaluated in processing. Adding the procssing of evaluating the match of date values into intervals for a significantly larger number of records can add overhead for procssing and join operations.

However, if you are dealing with a relatively small number of records (100, 200 thousand or less), then the impact will likely be relatively small.

So, if you need time dependency in modeling utlize them. As for the alternative modeling approach you mentioned, I have seen ODs modeling utilized moreso in cases where a complete history of master data, including all changes is required. I would not substitute the use of the delivered time-dependent MD attribute functionality with ODS modeling, but that could play a role in augmenting your approach.

Also consider that sometimes it can be useful to model both a time-dependent and non-time-dependent version of the same basic attribute, with slightly different names (and some user training on the difference).

For more on modeling options, see the how to paper on Enterprise Data Warehousing.

Best Regards -

Ron Silbestein,

SAP NetWeaver RIG, SAP Labs

Former Member
0 Kudos

Ron,

Thanks for your early response. We have relatively small number of customer records like 2000. We would go with delivered time dependent attribute functionality.

Regards,

Radha