Hey Guys, having a small renderfarm is cool for small and fast projects. Problem is that if you would like to scale this up in the future, you will hit the following problems:
1. Provisioning your computers with OS's and installing applications + plugins takes a LOT of time, major software install must be performed on each server if you don't have a good provisioning solution which will make it easy to create an image of one of your machines and install that on the other machines.
2. Electricity upgrade, your computers will begin using lots of electricity when they are rendering, and you must have a good installation so you avoid any cable meltings or blown fuses.
3. You will begin doing server administration instead of focusing on your projects, which I believe will lead into loosing more money than actually getting by using your servers.
4. Heat issues - you will have to ventilate the room where you keep the servers properly, AC could also be needed.
5. Licensing, you will have to purchase or rent licenses for all your nodes.
6. Networking bottleneck - this happens if you have 20 servers plus I guess, or very big scenes and your 1GB connection won't be enough and you need to go on 10 GB Fiber Optics, which ad up to the costs a lot.
7. Your project assets must have proper paths so all your nodes can find the assets, otherwise you will have missing textures or not properly set outputs, this can be solved by only working on your network storage I guess and using network paths.
8. If rendering big res images, all your nodes must have enough ram, like 48 - 96 GB if you are doing 8-12 K res. Having multiple nodes with poor performance won't do anything on these monster res images.
It can be a fun project, but I would advise giving our farm a test. We offer $20 worth of credits, enough for you to make an impression. Here's a quick video on how our process works.
https://www.cgifarm.com/renderfarm-quick-start-guideGood luck with your project guys! My project also begin like building something small for my studio, but I ended up focusing only on the renderfarm in the end, it had to be one or the other :).