You see this with any service as it scales up, that's where the learning occurs. As the number of requests climbs you reach limits, those limitations manifes

You see this with any service as it scales up, that's where the learning occurs. As the number of requests climbs you reach limits, those limitations manifest to the end-user as suckage, slowness, errors, and in the worst cases lost data. You can't plan those limits out ahead of time, you discover them as you go.