Provably Efficient Service Function Chain Embedding and Protection in Edge Networks

• Published in: IEEE Transactions on Networking Vol. 33; no. 1; pp. 178 - 193
• Main Authors: Zheng, Danyang, Cao, Xiaojun
• Format: Journal Article
• Language: English
• Published: IEEE 01.02.2025
• Subjects: Bandwidth; dedicated protection; exact algorithm; Firewalls (computing); Heuristic algorithms; Monitoring; Network function virtualization; Object recognition; Protection; Reliability; Reliability theory; reliable service function chain; Service function chaining; Smart manufacturing
• ISSN: 2998-4157; 2998-4157
• DOI: 10.1109/TNET.2024.3475248

Internet-connected devices generate service function chain (SFC) requests for reliability-sensitive applications such as smart factories and intelligent healthcare. To facilitate reliable SFC provisioning, one can employ dedicated SFC protection approaches to protect the primary service function path (SFP) by constructing a fault-disjointed backup SFP. Notably, the construction processes of the fault-disjointed primary and backup SFPs are interdependent, and the direct application of existing SFC embedding approaches to construct these SFPs by separate processes may not effectively optimize overall resource consumption. In this work, for the first time, we comprehensively study how to embed and protect an SFC through collaborative processes that have provable bounds. We formally define the problem of SFC embedding and protection (SFCEP), for which we develop the novel techniques of a backup SFP identifier (BSI) and resource-aware Bellman-Ford loop (RBL) to address the challenges posed by collaborative embedding and protection. On the basis of these techniques, we propose an efficient algorithm called optimal SFC embedding and protection (Opt-SEP). When the network resources are sufficient to accommodate an incoming SFC request, we prove that Opt-SEP can minimize the overall resource cost of creating a pair of fault-disjointed primary and backup SFPs. Moreover, for cases in which the network resources are limited, our extensive simulation results show that Opt-SEP significantly outperforms the benchmarks.