The following list includes the 8 most common causes of poor performance in a Qlik solution:

​

1. Dedicated hardware:
   
   If hardware is distributed among virtual hosts, it can lead to unstable performance. Instead, hardware should be dedicated to each host.
   
   Ensure that Qlik servers are not moved to slow hardware and that they do not end up competing for resources with other guests on the host.
    

2. The importance of CPU vs RAM:
   
   If you are experiencing performance issues with QV, it is often due to a RAM problem rather than limited CPU. However, if RAM usage is only around 60% and there are no errors, the problem is likely a weak CPU.
    

   Check that Hyper-threading and Turbo Boost are enabled to optimize CPU utilization.
    

3. SAN, not NAS:
   
   It is recommended not to use a file server for storing Qlik files. However, if you wish to use a file server, it should be SAN (Storage Area Network). NAS (Network Attached Storage) is not supported.
    

4. If performance logs indicate RAM problems, you can take the following steps:
   
    >  QV cannot reuse old cache.
   Schedule a service restart every night.
   
    >  Design of a specific QV app causing high resource usage:
   Inspect each app and examine the data model and charts. Reports with multiple dimensions and/or expressions should have conditions set for calculations. Otherwise, a single report can consume all available RAM if opened without selections.
   
    >  Run QVS Cleaning:
   https://help.qlik.com/en-US/qlikview/12.1/Subsystems/Server/Content/QlikView%20Server/QVSRM_Cleaning_Shared_Files.htm
    

5. QMC Settings (Qlik Management Console):
   
   There are certain settings in the QMC that can affect Qlik server performance.
    

   >  Decrease session and document timeouts on the server.
   >  Set the "working set" to 85-95% to optimize RAM utilization.
    

6. Follow Qlik's Best Practice for VM-ware:
   
   >  https://community.qlik.com/docs/DOC-2362
   
   >  NUMA:
   
   If NUMA (Non-Uniform Memory Access) is enabled, it can cause a decrease in performance. In some environments, this has no impact, but it can affect performance by up to 10-15%. The QV Server Service is more affected than the Publisher since the former relies more on fast cache access.
   
   >  If hyper-threading is enabled, the CPU will double the logical process.
   >  If Turbo Boost is enabled, the speed will increase. Use hardware approved by Qlik.
   
    

7. Use hardware approved by Qlik. 
   
   Examples of approved CPUs:

   • Intel® Xeon® Processor E5-2699 v4, 3.60 GHz (with turbo boost)

   • Intel Xeon E5-2679 v4 @ 2.50GHz

   • Intel Xeon E5-2699 v3 @ 2.30GHz

   • Intel Xeon E5-2690 v4 @ 2.60GHz

   • Intel Xeon E5-2697 v3 @ 2.60GHz

   • (22 cores, 44 with hyper-threading)
      

8. QlikView Server Logs:
   
   The best way to troubleshoot and address issues with the QlikView Server (QVS) is through the following logs. We recommend installing the QlikView application 'Governance Dashboard' which provides a clear overview of the logs.
    

   • Session Log:
     Displays user sessions. A session is defined as one user connected to one document.
      

   • Event Log:
     Updated every time QVS introduces a log entry in the Windows event log
     
     When troubleshooting, focus on the first error message.
      

   • Performance Log:
     Shows server performance metrics such as CPU and RAM usage. It also indicates server restart times.
      

   • Audit Log:
     The end-user audit log contains information about user selections, including sheet activation, bookmark usage, accessing reports, and maximizing objects. Not all customers have enabled the Audit Log. Audit Logs can become large because they contain all user actions in the documents."