vix.ing · top · new · best · stats · spec

Right buffer sizing matters: some dynamical and statistical studies on\n Compound TCP

2016/04/19 by Debayani Ghosh, Ghosh, Debayani, Krishna Jagannathan +3
Business, Management and Accounting · Computer Science · Engineering · #Advanced Queuing Theory Analysis #Advanced Wireless Network Optimization #FOS: Computer and information sciences #FOS: Electrical engineering #Network Traffic and Congestion Control #Networking and Internet Architecture (cs.NI) #Performance (cs.PF) #Systems and Control (eess.SY) #electronic engineering #information engineering

paper · pdf · doi:10.48550/arxiv.1604.05516

openalex publication_date 2016/04/19 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Motivated by recent concerns that queuing delays in the Internet are on the\nrise, we conduct a performance evaluation of Compound TCP (C-TCP) in two\ntopologies: a single bottleneck and a multi-bottleneck topology, under\ndifferent traffic scenarios. The first topology consists of a single bottleneck\nrouter, and the second consists of two distinct sets of TCP flows, regulated by\ntwo edge routers, feeding into a common core router. We focus on some dynamical\nand statistical properties of the underlying system. From a dynamical\nperspective, we develop fluid models in a regime wherein the number of flows is\nlarge, bandwidth-delay product is high, buffers are dimensioned small\n(independent of the bandwidth-delay product) and routers deploy a Drop-Tail\nqueue policy. A detailed local stability analysis for these models yields the\nfollowing key insight: smaller buffers favour stability. Additionally, we\nhighlight that larger buffers, in addition to increasing latency, are prone to\ninducing limit cycles in the system dynamics, via a Hopf bifurcation. These\nlimit cycles in turn cause synchronisation among the TCP flows, and also result\nin a loss of link utilisation. For the topologies considered, we also\nempirically analyse some statistical properties of the bottleneck queues. These\nstatistical analyses serve to validate an important modelling assumption: that\nin the regime considered, each bottleneck queue may be approximated as either\nan M/M/1/B or an M/D/1/B queue. This immediately makes the modelling\nperspective attractive and the analysis tractable. Finally, we show that\nsmaller buffers, in addition to ensuring stability and low latency, would also\nyield fairly good system performance, in terms of throughput and flow\ncompletion times.\n

Related