The purpose of this paper is to assess the impact of timer-based burst assembly algorithms for TCP traffic. We present an analysis for short, medium and long assembly times and investigate segment and flow distribution over the assembled bursts. Further, we also analyze their impact on the congestion window evolution and on the effective throughput achieved. It has been found out that short assembly times are ideally suitable for sources with small congestion windows, allowing for a speed up, while large assembly times yield a lower throughput variation among the individual assembled flows. For long assembly times, the transfer of more segments from the same source is trading off the increase of the burstification delay but no throughput gain is obtained. However, large assembly times smooth out individual flow performance and provide a significant lower variation of throughput. To this end, in this paper, we propose a new adaptive burst assembly algorithm that dynamically assigns flows to different burstifiers based on their instant window size.
TCP traffic analysis for timer-based burstifiers in OBS networks / K. Ramantas; K. Vlachos; O. Gonzalez de Dios; C. Raffaelli. - STAMPA. - 4534:(2007), pp. 176-185. (Intervento presentato al convegno ONDM 2007 - Optical Network Design and Modeling tenutosi a Atene nel Maggio 2007).
TCP traffic analysis for timer-based burstifiers in OBS networks
RAFFAELLI, CARLA
2007
Abstract
The purpose of this paper is to assess the impact of timer-based burst assembly algorithms for TCP traffic. We present an analysis for short, medium and long assembly times and investigate segment and flow distribution over the assembled bursts. Further, we also analyze their impact on the congestion window evolution and on the effective throughput achieved. It has been found out that short assembly times are ideally suitable for sources with small congestion windows, allowing for a speed up, while large assembly times yield a lower throughput variation among the individual assembled flows. For long assembly times, the transfer of more segments from the same source is trading off the increase of the burstification delay but no throughput gain is obtained. However, large assembly times smooth out individual flow performance and provide a significant lower variation of throughput. To this end, in this paper, we propose a new adaptive burst assembly algorithm that dynamically assigns flows to different burstifiers based on their instant window size.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.