Empirical study of buffer management scheme for Diffserv assured forwarding PHB

Abstract

-Based on an empirical study, this paper makes the following key contributions: Firstly, the results show that for ON-OFF traffic, RIO is better than WRED in protecting packets marked for treatment with lower drop precedence. Secondly, for shortlived flows, RIO achieves higher transactional rates than WRED. Thirdly, for bulk transfer, RIO and WRED achieve comparable long-term throughput. Finally, this paper also reports the results of experiments with 3 different models for setting of WRED and RIO parameters. We recommend the "staggered" model as best suited to achieve the requirements of the AF PHB.

Modeling of TCP behavior in a DiffServ network supporting assured forwarding PHB

“…They are shown in Table I. It can be noted that we chose a staggered way of WRED parameter setting [6]. The output queue size Q max is set to 50 packets.…”

“…It makes it easy to develop a model of a system consisting of various types of DiffServ node elements (meters/markers, shaper/droppers) just by changing some of the equations. We present sample analysis for a DiffServ domain with single rate three color markers (srTCM) [5] at edge routers and WRED queue (variant of Multi-RED queue [6]) at core routers. Note that we do not make an effort to find a configuration assuring that traffic aggregates achieve their target rates.…”

Improving the quality and reliability of traffic differentiation in IP networks

“…8 Excess rate is used to denote the difference R in i − R max i , if R in i > R max i , and is represented by excess i . 9 Due to the high number of possible differentiation schemes only specific configuration modes are covered. The selected examples were taken from a scenario where three classes contend for a 4.5Mbps link, with packet sizes uniformly distributed over [250,750] bytes.…”

“…This means that the node buffering resources are shared by all traffic classes and, as they are mapped to independent queues, the corresponding queue sizes may vary dynamically during the node operation. This coupled operation mode is different from traditional AQM techniques, such as RIO-coupled [9], where the dropping decision is centered on a particular traffic class. With this purpose the loss differentiation module has to evaluate, for each traffic class, a priority value reflecting the likelihood of packet dropping, i.e.…”

Applying adaptive virtual queue to improve the performance of the assured forwarding service

“…DAG02/03.EG20) priority of traffic indifferent virtual queues is same. WRED in [12] is an example of having only virtual queue with multiple drop precedence level. RIO is similar to FRED.…”