Tradeoffs in designing networks with end-to-end statistical QoS guarantees

Abstract

Recent research on statistical multiplexing has provided many new insights into the achievable multiplexing gain in QoS networks, however, generally only in terms of the gain experienced at a single switch. Evaluating the statistical multiplexing gain in a general network remains a difficult challenge. In this paper we describe two distinct network designs for statistical end-to-end delay guarantees, referred to as class-level aggregation and path-level aggregation, and compare the achievable statistical multiplexing gain. Each of the designs presents a particular trade-off between the attainable statistical multiplexing gain and the ability to support delay guarantees. The key characteristic of both designs is that they do not require, and instead, intentionally avoid, consideration of the correlation between flows at multiplexing points inside the network. Numerical examples are presented for a comparison of the two designs. The presented class-level aggregation design is shown to yield very high achievable link utilizations while simultaneously achieving desired statistical guarantees on delay.

Statistical service assurances for traffic scheduling algorithms

“…The results in this paper can also be applied to rate-based scheduling algorithms 39 . If such algorithms are rate-proportional, the derivations of statistical bounds with e ective e n velopes are quite simple 24 .…”

Toward statistical QoS guarantees in a differentiated services network

“…Providing statistical services within the intserv model (C and D) Much progress has been made in providing statistical services under the intserv model both in the singlenode [5,15] and multiple-node [17,28] cases. Most of these studies focus on a single-node first, and then extend to the multi-node scenario [15,17].…”

“…Most of these studies focus on a single-node first, and then extend to the multi-node scenario [15,17]. Statistical service has been investigated using envelopes of bounding moment generating functions [5].…”

“…A number of projects ( [5,15]) address the problem associated with Vertex C in Fig. 1, while studies of [2,7,13,17,18,28] are for the case of Vertex D. For the cases related to Vertex F, solutions have been discussed in [23,24]. The problems associated with statistical services within the diffserv model seem to be particularly challenging.…”

<title>Economics of scalable network services</title>

“…In reliable multicast, the use of forward error correction (FEC) techniques may require the initial transmission of additional, redundant data, but is more scalable than the traditional ARQ (automatic repeat request) technique of data repair used in unicast TCP. In QoS, there is a similar tradeoff between scalability and achievable resource utilization [43]. For example, reservation aggregation [24] can be used to improve the scalability of intserv, but at the potential cost of reduced resource utilization [25].…”