We present a reflector service that seeks to maintain application-level connectivity in the presence of network-level multicast failures. The service is based on the dynamic deployment of autonomous reflectors modelled as mobile agents on top of an active network infrastructure. It is able to repair multicast tree failures by building a self-organising tree of reflectors, which will be connected to each other via unicast. The scheme is decentralised and takes into account node and link resources to find agent locations that lead to low cost tree configurations. We focus on the basic decision mechanisms related to code mobility during the tree construction and destruction phases, namely: cloning, migration, merging and termination. We show some preliminary simulation results that confirm the viability of the approach and settle directions for further research. 1

The current paradigm of Content Distribution Networks (CDN) deployment requires heavy infrastructure investment since a large number of servers have to be deployed over a wide area. To overcome this difficulty, we propose a new paradigm where future CDNs are to be deployed over a leased server infrastructure. This paradigm shift changes the object placement problem in current CDNs to a combined object placement and resource provisioning problem. In this paper, we present a formulation of this problem, which is a combinatorial optimization, and prove its NP-hardness. We also present a heuristic solution to this problem based on Lagrangian relaxation and subgradient optimization, which also provides error bounds so that the accuracy of the obtained solution can be evaluated. The algorithm has been implemented and tested on a number of randomly generated network topologies, and results have shown promising solution accuracy.

Current Content Distribution Networks (CDN) deployment requires heavy infrastructure investment since a large number of servers have to be deployed over a wide area. This paper proposes a new paradigm where future CDNs are to be deployed over a leased server and network infrastructure. While this paradigm shift introduces a new dimension of flexibility, it requires the resource provisioning and object placement problems to be jointly considered. This paper formulates this new optimization problem, and presents solution to it based on Lagrangian relaxation and greedy heuristics.

Active and programmable networks allow innovative new services to be deployed rapidly. However, in a carrier grade network, it is imperative to maintain a scalable fast path mechanism so that the delay and throughput requirements are met. This is particularly important since in many cases network-level processing is only needed for a subset of packets and the remainder of traffic must be forwarded on the fast path. It is a challenge to design a cost effective node architecture that can satisfy this requirement. Current models are often `revolutionary' and may not scale to the required performance levels of a carrier grade network. In this paper we develop an evolutionary architecture that uses MPLS label stacks to enable efficient and scalable extraction of active packets from the fast path without penalising the rest of the traffic. Our model can be used to extend the reach of programmable networks over legacy infrastructure and provide a migratory path for existing carriers towards a programmable infrastructure.