First of all, I would like to congratulate you on the solution.
Another important point, I’m not DevOps, I know the basics of server administration.
I am responsible for a Saas solution, which is growing.
The scenario is that between 19 and 21 hours is used by customers. Out of these hours and little used.
Having said that, in my view it is not interesting to simply increase the number of CPUs.
I thought about separating web for a server, static files for another and database for a third server. However, it comes up against this issue of being used 3 hours a day again.
The best way I believe is using load balancer. So, as the system reaches 70% of its use, a new machine/cpu goes up.
Today I’m using Google Cloud Platform.
As much information as possible is welcome.