CPU spike is primarily due to the computation involved in QlikView. 10-15 minutes for the qvw to open in your server really concerns me a lot. You should definitely think about data model and UI optimization by understanding the application usage by the users. This is more of a strategic solution and will take more time.
However from the tactical standpoint, you can try the following things to get rid of the Exception.
1) Reduce the application with limited data based on the time period (or calendar period)
2) Split the fact tables based on the subject area instead of having 800million rows in one or two fact tables (as captured in the data model. Sorry the data model pic is not clear).
3) Break the application into various subject area and chain them using document chaining
Hi Manoj Kumar
Thanks again for the advice above. We figured out the issue was caused by a number of factors.
1) NUMA on our Servers was enabled which caused the Sever Instability
2) I modified the Application in question to amalagnate the tables into one major fact table.
If you have any queries with the above I would be more than happy to help.
Thanks very much.