摘要 |
<p>For each VP, a separate dynamic buffer is set up for each shaped VC and unshaped VC within the VP. The shaped cells stored in dynamic buffers are dequeued via VC scheduling (136) to a high priority queue (139) according to the shaped VC contracts, and the unshaped cells stored in dynamic buffers are dequeued in a round robin (142) manner to a low priority queue. The outputs of both the high priority queue and the low priority queue are passed to a second stage shaper where cells from the high priority queue are scheduled according to the VP contract, and cells from the low priority are also scheduled according to the VP contract, but only when VP bandwidth is not being used by the high priority cells.</p> |