A family of Applications: Hashing and Load Balancing

In this lecture, we will see our first glimpse of a “killer app” of probability in EECS. Probability helps us understand quantitatively how much of a resource we need when demands are somewhat random. There is a wide range of applications, especially in networking where understanding randomness is the key to the idea of “stochastic multiplexing.” The basic idea behind this (which are explored a lot more in EECS122 and EECS126) is that even though the universe of potential demand is vast, in the real world, the actual demands are random and not all potential “users” of the resource will show up at once. Consequently, we can “overbook” and share the resource while actually not provisioning for the worst-case total demand.