F2Scaling up vs scaling out

The problem

1 s on screen = 100 ms

Your one server is big enough. What happens the day it goes down?

One bigger machine

One server with eight workers handles 40 requests a second, with room to spare.

Live metrics for each version of the scene
Metric	Scale up	Scale out	Change
	—	—	—
target ≤ 400 ms	—	—	—
	0/s	0/s	—

F2Scaling up vs scaling out

When one server isn't enough you can buy a bigger one or add more of them. Only one of those options survives a crash.