We describe a new methodology for analyzing optical buffers with nonuniform
distribution of the delay lines, the nondegenerate buffers. Coupled with a
n intelligent scheduling algorithm, it features improved system performance
leading to more efficient and larger optical buffers. We demonstrate that
the concept under bursty, both time-slotted and asynchronous traffic condit
ions, can lead to a sizable reduction in buffer depth requirements compared
to conventional solutions.