Improve Qlik Sense Enterprise Performance

[Anonymous]

our users are currently complaining about slowness in data loading, and slow responsiveness when they create and view dashboards in Qlik Sense Enterprise. 

Qns: how should we deploy 4 Windows 2012R2 servers to have the best Qlik Sense Enterprise Performance so the user experience is more smooth? 

Our tries:

1. currently, we configured the 4 servers to 1 Central Node and 3 RIM Nodes with all features turned on. and all 4 are virtual proxy as well. not sure if this is the best configuration. Infront of this 4 servers is a reverse proxy server which hits the Central Node server directly. 

2. however, we saw here, the left side part of the diagram: https://help.qlik.com/en-US/sense/November2018/Subsystems/ManagementConsole/Content/Sense_QMC/link-v...

that the correct way seems to be to configure just 1 virtual proxy and load balance the 3 other nodes.

[Gysbert_Wassenaar]

It depends (of course).  Load balancing is a very simple round-robin proces. That will result in all the nodes loading all the apps into memory. If you one very large app then you might want to that app loaded and accessed on only one node. That leaves more resources for the other apps on the other nodes. So it depends on how many apps you have, how big they are, how many users access them. What the access patterns of the users and the apps are.

[Anonymous]

Thank you. What kind of 4 server configuration would you best recommend if:
- about 1000 users - all are given same access to create, share and view dashboards.
- users are starters of Qlik so their app is not complicated and not heavy.

what would be a generic config that will allow them to experience the best of what 4 servers can provide them?

[Joan_MARTY_P3]

Hello are you in shared persistance ?

If so, main ressource consumptions are on the database and hard drives from the fileshare.

how fast are your hard drives ?

[Anonymous]

Yes, the 4 servers (by the way, they are VMs) is configured to be shared persistence.
Hard drives are usual, not SSDs.

[Joan_MARTY_P3]

Have you checked in the perf log if your hard drives have an high latency?

Also have you optimized your postgres database using a bit more of shared memory for example ?

Next point should also be about your overall set up, have you dedicated your nodes depending on the usage ?

[Anonymous]

Hi Joan, what would be your design for the overall setup?