No Rest for the BI Admin (Continuous Improvement)

by Matt Christensen

May 11, 2010

I keep telling myself (and family members) that things are about to get better.  Business users will get the data they need.  Reporting tools will enable self-service.  Project schedules will lighten up. 

Then reality sets in.  Unfortunately there is a never-ending todo list for BI administrators and consultants.  Every new business initiative contains a BI component.  There is constantly an opportunity to provide information in a better format, more quickly, or more frequently.

In those rare moments where my todo list is short enough, there are a few common things I like to check in the BW system to ensure it is remaining healthy.  Although project demands are intense, it is important to the overall organization to keep an eye on these things even after go-live.

The first place I look is at the BW-specific database consumption.  I'll use transaction code db02 or similar depending upon authorizations.  Within the analysis menu, I can then find a BW-specific overview.  I drill into areas of this analysis where table sizes seem large to find the specific offenders.  Sometimes I will find a particular DataSource where the PSA hasn't been cleaned up recently.  My preference is to add a step to the process chain that loads this DataSource to delete it after a certain number of days -- perhaps 10-30 depending upon the load specifics.

Another quick analysis can be found in t-code rsbatch under the background jobs button.  Here I can see the utilization of background processes throughout the hours of a day.  I look for times when we are over and underutilizing the system and drill into the details.  Perhaps a job can be shifted by an hour or two to enable more smooth performance.  In other cases I might find a full load that has grown large and needs to be converted to a delta load somehow.

Finally, I always feel analyzing (SAP Business Content) BI statistics is a meaningful use of my time.  There are both query and data load statistics at various levels of detail.  I like to aggregate these results in either a BEx query or by using t-code listcube to look for the worst performers and/or patterns of degrading performance.  This might lead me to make changes to a query, MultiProvider, or even an underlying data provider or load process.

Even though we're constantly consumed with new project tasks, keeping an eye on these key areas of the system will help our organizations to efficiently use hardware and software investments.

An email has been sent to:

More from SAPinsider


Please log in to post a comment.

No comments have been submitted on this article. Be the first to comment!