Back-of-the-Envelope Estimation

Before you choose a single part, a few lines of arithmetic tell you whether the system needs one server or a hundred. You will learn to turn daily users into requests a second, to size for the peak with a spare rather than for the average, to estimate storage over a retention window, to turn an availability target into a downtime budget, and to put the common latencies in order. These numbers are what make a requirement such as "handles our traffic" checkable, and what later topics test their designs against. Precise capacity planning and load testing are left out: the aim is the right order of magnitude, from assumptions you write down.