The Web Replica Allocation and Topology Assignment Problem in Wide Area Networks: Algorithms and Computational Results

Computational Science – ICCS 2006

2006

Abstract. In this work we address the problem of anycast flow assignment. Anycast is a one-to-one-of-many delivery technique that allows a client to choose a content server of a set of replicated servers. We formulate an optimization problem of anycast flows assignment in a connection-oriented network, which is 0/1 and NP-complete. Thus, we propose a new effective heuristic algorithm based on Lagrangean relaxation technique. To our best survey, this is the first study that applies the Lagrangean relaxation to anycast flow problem. We evaluate the performance of the proposed scheme by making a comparison with its counterpart using a sample network topology and different scenarios of traffic demand patterns and replica location. Obtained results show advantage of the Lagrangean heuristic over a previously proposed algorithm.

Section: Resultsmentioning

confidence: 94%

Section: Resultsmentioning

confidence: 99%

See 1 more Smart Citation

Lagrangean Heuristic for Anycast Flow Assignment in Connection-Oriented Networks

Computational Science – ICCS 2006

2006

“…The problem of replica server location has been addressed in many works, e.g., in [16][17][18][19][20][21][22][23]. However, most of papers focus on well-known optimisation problems: the kfacility location problem and the k-median problem (see e.g.…”

Section: Review Of Related Workmentioning

confidence: 99%

Simultaneous optimization of unicast and anycast flows and replica location in survivable optical networks

Rak

2011

Telecommun Syst

Most of previous works related to survivable networks have been focused on unicast communications. In this paper, we address the problem of network survivability in the context of anycast communications, which has not attracted much attention recently. Anycast transmission is defined as one-to-one-of-many. To assure the network survivability, we use the single backup path approach, i.e., each demand has two node-disjoint paths to provide protection against a single link failure. We formulate new ILP models to find the optimal paths for anycast and unicast connections, as well as to find the optimal location of replica servers. To test the proposed approach, we run extensive numerical experiments using CPLEX solver on four example network topologies with different scenarios of replica server count and the proportion between unicast and anycast traffic. Results prove the efficiency of our solution in terms of various parameters including cost, path length, resource utilization, and time of computation.

“…One of exemplary techniques that apply anycast traffic is Content Delivery Network (CDN). For more details on anycast flows and CDNs refer to [3], [8][9][10]. In this work we consider on anycast unsplittable flows.…”

Section: Introductionmentioning

confidence: 99%

Unsplittable Anycast Flow Problem: Formulation and Algorithms

Computational Science – ICCS 2006

2006

Abstract. Our discussion in this article centers on a new optimization problem called unsplittable anycast flow problem (UAFP). We are given a directed network with arc capacities and a set of anycast requests. Anycast is a one-to-oneof-many delivery technique that allows a client to choose a content server of a set of replicated servers. In the context of unsplittable flows, anycast request consists of two connections: upstream (from the client to the server) and the downstream (in the opposite direction). The objective of UAFP is to find a subset of the requests of maximum total demand for which upstream and downstream connection uses only one path and the capacity constraint is satisfied. To our best survey, this is the first study that addresses the UFP (unsplittable flow problem) in the context of anycast flows. After formulation of UAFP, we propose several heuristics to solve that problem. Next, we present results of simulation evaluation of these algorithms.