SUMMARYRecently, there has been an increasing research interest being directed to optical burst switching (OBS) networks because of its mature electronic processing capability and high-capacity optical transport capability. The blocking time for an OBS node, representing the time duration in which a certain channel is occupied by some class of bursts, plays an important role in characterizing the performance of an OBS network. In this paper, we analyze this important metric for an OBS node with a horizonal and a single channel theoretically and numerically. On the basis of that, the burst arrivals are assumed to follow the Poisson process and the burst lengths have a phase-type distribution, we are able to formulate a working mechanism of an OBS node into a multilayer stochastic fluid model. Using a level crossing arguments method, we are able to obtain the Laplace-Stieltjes transforms of the blocking time, from which n-order moments and the probability density functions of the blocking time can be further derived. We provide a numerical example to illustrate the theoretical analyses, and we further discuss the effects of the offset time and the arrival density on the blocking time through this numerical example.