Hi Dear QV Community
I Made some tests and i found the following results:
1) Qlik NPrinting connected to a Local File on Qlik NP Server (Local Mode): When the reload meta data and publish task execution are running the resources (CPU and RAM) to open and navigate QVW File are consumed by the Qlik NP server by QlikView Desktop
2) Qlik NPrinting connected to a QlikView File located on a QlikView Server (Server Mode): When the reload meta data task are runing the QlikView Resources (CPU and RAM) are consumed by the server, but when the publish task are runing the resources (CPU and RAM) to open and navigate QVW File are consumed by the Qlik NP server by QlikView Desktop.
The scenario 3 could not be tested because the cluster installation are not ready.
My big concern is that the QlikView resources have to be consumed by the Qlik NPriting machine, this generates too slow the generation of reports when many users
your findings are correct. Cluster mode is pretty much the same as Server one.
Regarding your concerns, you have to consider that NPrinting potentially open up an equal number of qv.exe instances than the number of NPrinting Engine machine cores. So, potentially the amount of RAM consumed by NPrinting will be:
["single qvw RAM" x N° of cores]
This is why you should consider to keep number of cores under a reasonable number (the recommendation is to stay under 12 cores). Over this limit I strongly recommend to scale with the number of NPrinting engines.
Let me know if this helps.