Our client is planning on growing in the use of Qlik Sense, and we are pretty sure we will have to go to a multinode setup at some point.
Does anyone know of (I couldn't find any) some guide, article, discussion or documentation concerning when it is needed to move to a multinode setup? I mean some place where I can look at and say "Ok, we are going to have N concurrent users, and each one of them consuming an app of M Gb, so we need... (for example) 2-node setup of 36 Gb RAM each one".
If not, how can I correctly estimate the setup needed??
You can serve the same apps to users either, using a single big node or multiple smaller node. Qlik Sense is able to scale in as well as scale out. What is really matter is the total hw resources and the biggest application fitting at least one node.
Generally a multi node environment is more resilient to the failure but requires at least a good network between nodes.
You can check app memory footprint using windows monitoring tools.