Answers are grounded in this paper and cite their sources.
The performances of two important cell loss priority mechanisms (CLPM) in shared buffers withNservers, Partial Buffer Sharing (N‐PBS) and Push‐Out Buffer (JV‐POB), are compared. Each arriving flowiis guided to a shared queue of sizeKwhere it is served by the correspondingi‐th server. Further, these flows are Poissonean, independent of each other and consist of a uniformly distributed mix of high (H) and low (L) priorities. Subject to the required cell loss ratios for both priorities,clr*Landclr*H, we determine the maximum allowable traffic intensity λmaxfor each incoming traffic flow as a function of the global priority mix in the buffer of sizeK.For an arbitrary CLPM, the maximum allowable load is achieved in the symmetrical traffic situation where allNlinks carry the same load. Only when N> 1, theN‐POB scheme is found to be superior in performance toN‐PBSover the whole priority mix range and for realistic cell loss ratio requirements. Moreover, the difference in performance betweenN‐POB andN‐PBS increases with increasingN, but decreases with increasingK.Finally, a methodology based on buffer size scaling has enabled the computation of the performance of theN‐POB.
No takes yet. Share an insight, caveat, or question.
Mieghem et al. (1998) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: