This note evaluates the approximation of overflow probability put forward by Sakasegawa et al (in a paper published recently in Management Science [5]), and its level of accuracy when applied to traffic generated by bursty sources.