We have just started using QlikView against SAP BI. We are having some success, but are finding that the queries are taking a long time. We have found that turning off slicing will work, but know that that might not work with our larger queries. We are looking for any suggestions on what we can do to improve performance. Does anyone have any experience with performance issues when pulling data from BI?
I am also working with an accounting system and the loads of data can be cumbersome. The best thing to do is load your data into a QVD file which is an optimized Qlikview data file. You can then do an incremental load which will only load new data into the QVD file. This drastically increases the performance. Search incremental load on this site for more info.
Thanks for the response. We are planning to do incremental loads with our historical data. Our issue is our aging query, which pulls current aging for all inventory in the company. We can run the query from BEX just fine, but running through QlickView is taking too long.
By aging I assume you are talking about some kind of depreciation of your inventory and therefore you will need to reload all of the inventory data everytime. I'm not sure what BEX is but if the query in qlikview is slow you might try storing the query in the database as a view and then calling the query from qlikview. Not sure if this will help but it's worth a try.
I am assuming you are using OLAP connector to connect to your existing BEX query on top of the cube. I personally had lots of issues and limitations using that method and ended up going against cube using SAP connector and building star schema at QV side. Once you have this done, it runs pretty fast. So now I have Fact table, SID tables, Text tables and then attributes (as dimension) being downloaded priodically into QVDs and then all these are linked back in QV data model application. One good thing is by doing this, you can easily setup delta downloads from fact table and have QVD updated on daily basis using incremental loads. Let me know if you need more details. Lately, this seems to be what I am doing most of the time - dealing with BW cubes. :-)
Hope this help.
Last week I have done a seeing is believing using the OLAP connector to make use of a query. Due to the amount of records it isn't possible to load data by full loads so I need some Delta (incremental) method. Is this possibe using a query?
Stated above, I see incremental loads are possible when using the SQL connector that directly reads from cubes. Can you tell me in more detail how to set-up such a architecture? I've been searching the net for some description but I cannot find anything.
Thanks in advance!
We released a Service release last week with some performance improvements and in the Installation & Usage Guide there are some performance tricks mentioned, especially to split up the query in one for the Key Figures and one for each Charactersitics with the necessary attributes.
Regards, Lars Wahlstedt