Unlock a world of possibilities! Login now and discover the exclusive benefits awaiting you.
Hi all,
i have strange scenario at one of my customers.
We have setup a Qlikview Environment with a SBE Server:
- 32 GB RAM
- Multicore processor
We published a Qlikview App with a relative small amount of data, where the Fact Table has more less 1.000.000 of records.
There are SOME users when they performing a time based query in Qlikview, i.e. selection of multiple years, who have a severe performance issue, that means they have to wait between 15-30 sec. till the object and data are refreshed.
But launching the same query from other workstation's in Qlikview takes just 2 second of time. We are accessing it via Qlikview Ajax Client.
The server is not overloaded during the action:
- 10 Gigabyte of RAM are still available when we launch the query from that specific workstation
- Less CPU are used during the calculation
Somebody could give me a hint for further investigations I could make?
We have tried to use a laptop in that office with the same LAN connection and it works fast.
Are there some specific client requirements?
As far as the calculation and processing of the requests of the Client has be done by the Qlikview Server ....
Thanks,
Patric
You can add calculation conditions to charts, so that the chart does not calculate until the conditions are met.
e.g. Count of rows < 20000 or Count(Year) < 3, or the relevant chart variable is enabled.
The calculation condition is on the chart properties, General tab.
You can also customise the Error Message for the calculation condition.
Hidden and minimized charts are not calculated.
Use Document Analyzer (http://qlikviewcookbook.com/recipes/download-info/document-analyzer/) to identify which charts are taking longest to calculate. That will give some focus and objective measurements to your tuning efforts.
-Rob
Hi,
just for Info the Performance Problem were related to the complex Expressions I had in the chart.
I moved the logic from GUI to Script and now it is much better.
Thanks to all,
Patric