Skip to Main content Skip to Navigation
Conference papers

An Optimised Dynamic Resource Allocation Algorithm for Cloud's Backbone Network

Ilhem Fajjari 1, 2 Nadjib Ait Saadi Guy Pujolle 1 Hubert Zimmermann
1 Phare
LIP6 - Laboratoire d'Informatique de Paris 6
Abstract : Sky computing is a promising concept enabling a flexible deployment of geographical distributed applications. Whereas, it is faced with a fundamental challenge which is: “efficient resource utilisation” within Cloud's infrastructure. Hence, a high flexible and intelligent resource allocation scheme is necessary to accommodate unpredictable and variable users demands. This paper tackles the fundamental challenge of efficient resource allocation within Cloud's backbone network. The ultimate goal is to satisfy the Cloud's user requirements while maximising Cloud provider's revenue. The problem consists in embedding virtual networks within substrate infrastructure. A new dynamic adaptive virtual network resource allocation strategy named Backtracking-VNE is investigated to deal with the complexity of resource provisioning within Cloud network. The proposal coordinates virtual nodes and virtual links mapping stages to optimise resources usage. Moreover, thanks to forecasting module, Backtracking-VNE guarantees an efficient resources share between embedded virtual links with respect to their occupancy. We demonstrate through extensive simulations that contrarily to static bandwidth allocation approaches, Backtracking-VNE enhances substrate bandwidth usage whilst minimising virtual links congestion. Acceptance rate of virtual networks and Cloud providers income are also improved compared with related strategies.
Complete list of metadatas
Contributor : Farida Benaouda <>
Submitted on : Wednesday, January 1, 2014 - 8:40:31 PM
Last modification on : Friday, December 13, 2019 - 11:50:05 AM



Ilhem Fajjari, Nadjib Ait Saadi, Guy Pujolle, Hubert Zimmermann. An Optimised Dynamic Resource Allocation Algorithm for Cloud's Backbone Network. The IEEE Local Computer Networks (LCN), Oct 2012, Clearwater, United States. ⟨10.1109/LCN.2012.6423621⟩. ⟨hal-00923071⟩



Record views