Deployment & Management

[Yousef_Amarneh]

Hi,

I've installed 2 Qlik Sense servers (one central node and one rim node) and I want to configure the load balancing between the central and rim node, what I want to know, do I need to use network load balancer or this is out of the box feature in Qlik sense?

Also I need to know the steps how to configure the load balancing between the nodes and how to test it.

Thanks in advance.

[Levi_Turner]

Hey Yousef,

If you want to load balance authentication (QPS) then an external device / application will be needed to do this.

If you want to load balance app delivery then just use Virtual Proxies > Edit > Load Balancing to add multiple Engine resources.

Hope that helps.

[daiana_rossignol]

@Levi_Turner If you want the users to access only one url instead of the different node names or ips, do you need a network balancer or QlikSense can also do it? through virtual proxies?

Lets pretend you have 2 nodes primary (192.168.101.1) and secondary (192.168.101.2) and secondary works as failover candidate. I want users to access one a only one ip (192.168.101.3) so that when primary fails, then they access secondary, no need to change address. 

Can you do this without a network balancer? can you do it through virtual proxies?

Thanks ind advance

BR 

Dai

[Levi_Turner]

Hey @daiana_rossignol,

You will need to do it on the network level, most easily with a load balancer / reverse proxy.

In this work-flow:

1. User types in https://sense.company.com into the browser
2. The user's machine uses DNS to translate sense.company.com into an IP address
3. The browser then reaches out to that IP address on a given port (default for the HTTPS protocol is 443)
4. If the server has a process listening to incoming traffic on 443 then it will respond
5. ...

In the scenario you laid out, the difficulty will be that step (2), without handling on the network level) will translate to a single IP address.

The benefit of a network appliance is that it is bound to a single IP externally but translates that request and then routes to one or many servers to handle the request. Most network applications or appliances can do health checks on the pool of servers it routes to to ensure that they are actively responding.

I've heard tell of other approaches but since I am not a networking guru, I can't vet them.

As for Qlik Sense, there are two key components:

• Proxies
• Virtual Proxies

Proxies are webservers. They bind to a port. So the server at 192.1.1.1 is available at the ports configured for the Proxy service.

Virtual proxies are methods of authentications (with paths, called prefixes). This means that if you have a virtual proxy with the path of /saml attached to a proxy, then https://192.1.1.1/saml will respond but https://192.1.1.1/saml2 will not.

You can bind virtual proxies to multiple proxies, which is especially ideal for more complicated authentication mechanisms (e.g. SAML or JWT), but that is outside of the bounds of how users are routed into Qlik Sense.

Hope that helps.

[datamakery01]

@daiana_rossignol Beside your Primary and Secondary node. Do you have a third server that holds the repository database and shared folder that the 2 nodes can access? I was advised to setup the 3rd server by a consultant to achieve "..so that when primary fails, then they access secondary, no need to change address. .."

however, i didn't try it. so wanted to know who else has done it and how did it go?

[daiana_rossignol]

Hello @datamakery01

Repo and bd are in DFS. clustered windows servers both.

hope it helps!

Dai