BW performance

[Former Member]

Hi,

Hope that this is the correct forum to post this topic.

I have new project to improve the reporting that we currently got of BW using BEx. Our biggest issue is speed or the lack thereof. I am open to suggestions as to what areas I need to focus in on in order to troubleshoot slow running reports.

I should add that I am very new to BW and am at project manager level - not technical. However, I do not want to be completely at the mercy of the consultant that I have access to.

Your feedback is much appreciated.

Kind regards,

Regrads

[Former Member]

Hi ,

1. The best way to improve query performance - is to create aggregates.Aggregates are replica of cubes,with some data restrictions.Actually the data content of aggregates are lesser than cubes, so it's faster to retrive data from aggregates.

But before going for aggregates, you need to identify the KPI of your reports and the most occouring parameter values (that users are providing) for those KPIs.

This will help you to design your aggregates.

2. Except that , you can do compression of your cubes. but if you go for compression, you will never be able to delete your cube data request wise.

3. You can do indexing, this will also help to improve the performance to some extent.But, usually we used to do indexing after each data load to infocube.

4. Another one thing, we can do is partitioning, but for that also we need some specific infoobjects should be present in the info providers - eg - Fiscal Year etc.

Thanks,

Biswarup

[Former Member]

Hi ,

1. The best way to improve query performance - is to create aggregates.Aggregates are replica of cubes,with some data restrictions.Actually the data content of aggregates are lesser than cubes, so it's faster to retrive data from aggregates.

But before going for aggregates, you need to identify the KPI of your reports and the most occouring parameter values (that users are providing) for those KPIs.

This will help you to design your aggregates.

2. Except that , you can do compression of your cubes. but if you go for compression, you will never be able to delete your cube data request wise.

3. You can do indexing, this will also help to improve the performance to some extent.But, usually we used to do indexing after each data load to infocube.

4. Another one thing, we can do is partitioning, but for that also we need some specific infoobjects should be present in the info providers - eg - Fiscal Year etc.

Thanks,

Biswarup