How to derive sample size formula for A/B testing from scratch?

Simply math. Math is sometimes self-explaining : )

Let X_i and Y_i be the observed value of metrics that we are interested in in test and control group, X and Y be the average each group. Assume normality,