Cloudonomics: The Business Value of Cloud Computing, + Website

MAD about Being Normal

There are a couple of other scenarios worth touching on. One is a normal distribution. Let’s assume that we are playing tennis from the center of the court and the person on the other side of the court is actually a friend, just out to hit balls with us. She is a decent player but not an expert, so she tries to hit balls back to us but doesn’t always succeed. If the distribution of volleys is normal with a mean at the center of the court but with standard deviation σ, the expected distance that we will have to travel from the center is the mean absolute deviation (MAD)—which, for a normal distribution, is or roughly .798σ. If the distribution of demand is normal we can determine the likely penalty function: .798σ(½cd + ½cr). We expect to have to sidestep about four fifths of the standard deviation, and half the time it will be to the left, half the time to the right (i.e., if we pick capacity at the expected value of demand, half the time we will be under and half the time we will be over).