What is the problem to pay for bandwith and workload on the server?
We pay for the hosting of our saas and that everything is online - that bubble has a great software we can build our tools is just the reason why we choose you as provider. (Same like when you build a website or when you choose a hosting provider).
Of course we use bubble cause we love the software - but this is just the start.
I would do a base flat for capacity and bandwith (so even if you use different data storage systems - users have a workaround for storage limit right now - i pay for what the user does with my app.)
Btw: you can also let us pay per usage of ram → this could be possible about the calculation how much “Time” the server needs for the work he does for us - so the cluster has no capacity limit - i just pay for how long it took to calculate stuff… Online Rendering Clouds use this functionality.
So why should i need a “predictable” system?
If i rent a server the server has a limit my user can need - if i do not want this i go to amazon with my app - cause i care more that my app is online - then i do about some extra dollars.
Just show some examples how the usage + workflows running = time you pay on server.