Latency-Aware Backup Resource Optimization for Virtual Networks Against Single Facility Node Failure

Recently, the problem of how to efficiently back up virtual network function (VNF) computing resources to prevent single facility node failure has received widespread attention, where the computing resources required for all the VNFs activated in one facility node are assumed to be packaged and backed up to another one. However, in practical applications, VNFs are often allowed to be deployed in different facility nodes. Therefore, while considering that the computing resources required by VNFs are able to be backed up to different facility nodes, the problem of backing up the computing resources required by the VNFs of each facility node to other nodes in the network to prevent single facility node failure such that the latency requirement is satisfied and the total required computing resource consumption is minimized, termed latency-aware computing resource backup (LACRB) problem, is first proposed. This problem reflects realistic VNF deployment flexibility and addresses both latencysensitive protection requirements and overall resource efficiency. It provides a scalable and practical solution framework for network operators. To address this problem, we propose a linear programming-based randomized rounding (LPBRR) algorithm for efficient approximation. For solving the LACRB problem, we design an effective algorithm called dominating set-based resource protection (DSBRP) with a provable approximation ratio for the simplified version of the LACRB problem, and further extend it into an enhanced DSBRP (EDSBRP) algorithm for the general problem. Simulation results show that the EDSBRP algorithm achieves near-optimal results across various network sizes and failure scenarios while requiring substantially less computation time, demonstrating both its feasibility and efficiency under latency constraints.