In the case of the App Service, the amount of compute power that we need depends on the load distribution. For instance, if all the load on our functions came during a single hour, the service would need to scale out to a large number of instances during that hour. It is best to run a load test on your application to determine the load distribution and the resource consumption of the application. We can, however, simplify the calculation by making some reasonable assumptions.
For 100,000 user requests a day, if we assume an even spread of traffic load around the clock, the load amounts to 100,000/(24*60*60) = 1.16 requests a second, which can be easily handled by a single instance (remember that ...