[Paper Review] Survivable Probability of SDN-enabled Cloud Networking with Random Physical Link Failure

Zhili Zhou, Tachun Lin|arXiv (Cornell University)|Sep 10, 2017

Software-Defined Networks and 5G20 references3 citations

TL;DR

This paper introduces survivable probability as a new metric to evaluate the reliability of SDN-enabled cross-layer cloud networks under random physical link failures. It proposes the base protecting spanning tree set to exactly compute survivable probability without enumerating all cutsets, achieving optimal resilience with minimal shared links and outperforming lower-bound estimations from single spanning trees.

ABSTRACT

Software-driven cloud networking is a new paradigm in orchestrating physical resources (CPU, network bandwidth, energy, storage) allocated to network functions, services, and applications, which is commonly modeled as a cross-layer network. This model carries a physical network representing the physical infrastructure, a logical network showing demands, and logical-to-physical node/link mappings. In such networks, a single failure in the physical network may trigger cascading failures in the logical network and disable network services and connectivity. In this paper, we propose an evaluation metric, survivable probability, to evaluate the reliability of such networks under random physical link failure(s). We propose the concept of base protecting spanning tree and prove the necessary and sufficient conditions for its existence and relation to survivability. We then develop mathematical programming formulations for reliable cross-layer network routing design with the maximal reliable probability. Computation results demonstrate the viability of our approach.

Motivation & Objective

• To address the lack of exact, scalable reliability metrics for SDN-enabled cross-layer cloud networks under random physical link failures.
• To develop a method that computes survivable probability without enumerating all cross-layer cutsets, which is computationally infeasible for large networks.
• To design a resilient routing framework that maximizes survivable probability using minimal physical resource sharing.
• To provide a practical, exact solution for survivable network design in SDN environments, even when traditional survivable mappings do not exist.

Proposed method

• Introduces survivable probability as the exact probability that the logical network remains connected after any single physical link failure.
• Defines the base protecting spanning tree set as a minimal set of protecting spanning trees ensuring logical network connectivity under physical link failures.
• Develops mathematical programming formulations to compute the base protecting spanning tree set with the goal of maximizing survivable probability.
• Proves that maximizing survivable probability under uniform failure probabilities is equivalent to minimizing shared physical links in the protecting spanning tree set.
• Uses logarithmic transformation of failure probabilities to convert the non-linear survivable probability maximization into a linear integer program.
• Employs IBM ILOG CPLEX to solve formulations within 450-second time limits, demonstrating computational feasibility.

Experimental results

Research questions

• RQ1 How can survivable probability be exactly computed for cross-layer cloud networks under random physical link failures, avoiding approximation via failure polynomials?
• RQ2 What is the necessary and sufficient condition for the existence of a base protecting spanning tree set in a given cross-layer network?
• RQ3 Can the survivable probability be maximized without enumerating all cross-layer cutsets, and if so, how?
• RQ4 How does the proposed base protecting spanning tree set compare to a single maximal protecting spanning tree in terms of survivability estimation accuracy?
• RQ5 To what extent can the base protecting spanning tree set improve resilience in non-survivable cross-layer networks?

Key findings

• For survivable cross-layer networks like LN1-over-NSF, the base protecting spanning tree set achieves 100% survivable probability under both uniform and random physical link failure probabilities.
• In non-survivable networks such as LN2-over-NSF, the survivable probability increases monotonically as physical link failure probabilities decrease, with results computed within 1 minute per case.
• The maximal protecting spanning tree provides a lower bound estimation of survivable probability, and for failure probabilities below 15%, this lower bound exceeds 50% of the true survivable probability in all tested cases.
• The proposed method avoids the need to enumerate all cross-layer cutsets, requiring at most |EP| protecting spanning trees, which significantly reduces computational complexity.
• The base protecting spanning tree set enables an auxiliary protection scheme in SDI without introducing redundant physical resources, enhancing cloud service resilience.