Properties of the Traffic Output by a Leaky-Bucket Policer with Long-Range Dependent Input Traffic

Abstract

 Long-range dependence (LRD) is a largely verified property of Internet traffic, which severely affects queuing performance in network buffers. A common approach for guaranteeing performance requirements is to control the statistical profile of the input traffic by regulators based on the leaky bucket scheme. In this paper, we investigate by simulation how the 1/f α power-law spectrum of LRD traffic is altered when traffic is regulated by a leaky bucket policer. Analysis of the traffic spectral characteristics is carried out mainly by means of the Modified Allan Variance, a time-domain quantity with demonstrated superior accuracy in fractional-noise parameter estimation, recently introduced also for traffic analysis. This approach allows to get a finer insight into power-law spectral characteristics of policed traffic. We also investigate some other properties of the leaky bucket fed with LRD traffic, such as its dropping probability and its effect on queuing delay in a following FIFO scheduler.Index Terms  Communication system traffic, fractional noise, Internet, long-range dependence, queuing analysis, traffic control (communication).

Output Traffic Characterization of Policers and Shapers with Long-Range Dependent Input Traffic

“…In our previous paper [17], we presented a thorough simulation study, which confirmed that leaky-bucket policers can hardly weaken traffic LRD and that, consequently, it is difficult to match service delay bounds if α increases.…”

Queuing Performance of Long-Range Dependent Traffic Regulated by Token-Bucket Policers

“…This leads to suspect that providers might be unable of guaranteeing statistical delay bounds, if customers offer LRD traffic exhibiting large swings of α. We provided empirical evidence of this claim in [13] [14].…”

“…4 shows that there is no fGt R traffic series x S (t) with same parameters m xS = m xP , σ xS = σ xP that, for any value of α S , yields similar queuing delay distribution than policed traffic x P (t). In [13][14], we have shown that if the input traffic exhibits increasing LRD parameter α, still keeping same average rate and variance, policers deliver traffic downstream with α almost unaffected, causing possible disruptions of end-to-end delay SLA. While this is certainly true, now we can show also that the regulation action of the policer is however effective to improve the queuing performance in downstream schedulers.…”

“…In previous papers [13][14], we demonstrated by simulation that TB policers and shapers can hardly reduce traffic LRD (i.e., its α parameter) and that statistical bounds of network delay may not be met against increase of LRD of offered traffic. This leads to suspect that providers might be unable of guaranteeing statistical delay bounds, if customers offer LRD traffic exhibiting large swings of α.…”

On Traffic Long-Range Dependence at the Output of Schedulers with Multiple Service Classes

“…In a previous work, we showed by simulation that LRD cannot be reduced by standard leaky-bucket (LB) policers and shapers [7][8], unless by dropping a large fraction of traffic.…”