Before we move on, let's try to understand the characteristics of the system that we are going to use for our comparison. The central indicators for most web applications are throughput, latency, CPU, and memory usage. The web now has completely different requirements to when it started. Previously, computers were sequential. Users used to be happy to observe simple, static content, and the overall load on the system was low. The primary operations involved generating HTML or simple calculations. Computations fitted into one processor and we did not require more than one server for a web application.

Over time, the rules of the game have changed. The web started counting users by the billion, and content started being dynamic and even real-time. The requirements for throughput and latency have changed substantially. Our web applications have started to be highly distributed on cores and clusters. Understanding how to scale a web application has become critical. An important question is—how does the number of parallel workers change the latency or throughput?