I need to setup a system which will be used for hardcore animation using Adobe packages wherein 20 Animators will be able to work simultaneously. In order reduce capital and ramp up the max capacity of each system, I have thought of setting up a central mainframe server/processing unit (render farm - or maybe Nvidia grid) which will in turn power 20 computers.
Something similar has been done by LinusTechTips in which they demonstrated 10 Gamers 1 CPU.
This can be done in 2 ways:-
1) A super configuration central server powering 20 mid config workstations. In this case, designing will be done on 20 workstations while the render will be pushed to the backend server.
2) A super configuration central server powering 20 virtual PCs (Monitor, Keyboard & Mouse). In this case, designing as well as rendering will be done using the backend server. >> Preferred.
Q1) Are such a setups compatible with Adobe? If yes, then what kind of licensing does Adobe provide (per core or some other condition)?
Q2) Is GPU Compute ( Open CL or CUDA) supported on over the network render?
Q3) What kind of license would I need for the VMs?
Q4) What are the best practices while setting up such an architecture? Any guidance (forums, guides, etc...) will be highly appreciated.