Backup Resource Allocation Model for Virtual Networks with Probabilistic Protection against Multiple Facility Node Failures

Abstract

This paper proposes a backup resource allocation model for virtual networks with providing a probabilistic protection against multiple facility node failures by jointly considering backup computing and bandwidth resource allocation. Previous work only considers the backup computing resource allocation for virtual nodes regardless of the constraints from network aspects. In this paper, we firstly analyze the types of backup paths required for each virtual link in cases of multiple facility node failures. We describe the substrate link capacity constraints and flow constraints for backup paths, which show the interactions between backup computing and bandwidth resource allocation. In the proposed model, the routings of backup paths are determined when the backup computing resource allocation is determined. By adopting robust optimization with several extensive mathematical operations, we formulate the proposed model as a mixed integer linear programming problem to minimize the total required backup computing capacity. Finally, we show the optimal backup resource allocations for an example cloud provider with and without considering the constraints from network aspects, respectively.