In a Markov process, the elapsed time between arrivals is exponentially distributed. This simply means that short inter-arrival times are more probable than long ones. This is also known as a memoryless arrival pattern, because each arrival is statistically independent from the next.