Energy-aware traffic engineering with elastic demands and MMF bandwidth allocation

Abstract

In recent years, there has been a remarkable growth of the Internet energy consumption, which is expected to persist in the future at an even higher pace. At the same time the network access capacity of individual subscribers is rapidly reaching values high enough to move the traffic bottleneck from the access network to the core network in most scenarios. This will soon make the elastic nature of traffic an important aspect of network resource management and will require a redesign of the energy-aware traffic engineering techniques so far based on inelastic traffic demands. We propose a novel optimization approach to select a routing path for each elastic traffic demand and decide which routers and links to put to sleep so as to maximize a network utility measure depending on the traffic demand rates, while satisfying a constraint on the total energy consumption. Bandwidth is allocated to each elastic demand according to the Max-Min Fairness (MMF) paradigm, which approximates the resource allocation of the transport layer

Energy management in communication networks: a journey through modeling and optimization glasses

“…Some work deal with elastic consumption and surprisingly few consider multi period, survivability or special QoS constraints. [58] x x x x [59,60] x x x x x Smart protection [54] x x x x Robustness [61] x x x x x Robustness, Smart protection [41] x x Elastic traffic [62] x x Bi-objective function [24] x Network design decisions [21] x [22,[63][64][65][66][67] x [23,47] x x x Piece-wise concave energy costs [68] x x Bidirectional link capacity [69] x x [70] [71]…”

“…All the proposed approaches are suitable for offline centralized management mechanisms. In [41], a novel bi-level centralized optimization approach for EANM with elastic traffic demand (see the methodology can be applied offline only. To conclude the part dedicate to flow based routing, let us cite the very popular work of Chabarek et al [24], which is usually used as reference to prove the prevalence of the fixed consumption components in traditional network device.…”

Minimizing energy and link utilization in ISP backbone networks with multi-path routing: a bi-level approach

“…To this end, many technologies and approaches have been proposed by the networking community to cut the carbon footprint of the ISP backbone networks [1,3,7,10,14,16]. Among them, one promising solution is the energy management of network elements including routers and communication links [1,3,10,12]. In fact, network resources like processing power and memory are oversized in communication networks nowadays, which results in a low utilization of 30-40% in low traffic periods [14].…”

“…Following this idea, authors in [10] tried to minimize the total energy consumption by switching off unused network devices while guaranteeing full network connectivity. In [3], the authors studied the energy-aware traffic engineering problem subject to elastic traffic and max-min fair bandwidth allocation. The work in [1] is focused on the energy minimization strategy that selectively switches off devices according to the traffic level.…”

“…However, all the work mentioned above only considered the single-path routing (using protocols like Routing Information Protocol (RIP) or Open Shortest Path First (OSPF) protocol) for the energy-aware traffic engineering [1,3,10,12], where only one path is used per communication request to deliver data from its source node to its destination. The single-path routing strategy may lead to a high blocking probability when some network links are congested during the period of high traffic load.…”