[Paper Review] A Comparison of Approaches for Traffic Engineering in IP and MPLS Networks

Alan Gous, Arash Afrakhteh+1 more|arXiv (Cornell University)|Aug 12, 2016

Advanced Optical Network Technologies4 references3 citations

TL;DR

This paper compares traffic engineering (TE) techniques in IP and MPLS networks, evaluating their performance in operational settings. It analyzes constraint-based routing, RSVP-TE, and MPLS-TE, concluding that MPLS-TE offers superior control and predictability in real-world deployments, while IP-based TE faces scalability and consistency challenges.

ABSTRACT

As IP / MPLS network service providers drive for improved network efficiencies, traffic engineering is becoming increasingly important. In this article we review the approaches for traffic engineering in IP / MPLS networks and present the result of a study, which compares their performance in operational networks

Motivation & Objective

• To evaluate and compare the effectiveness of IP and MPLS-based traffic engineering approaches in real operational networks.
• To identify scalability, control, and performance limitations of IP-based TE mechanisms.
• To assess the practical deployment viability of constraint-based routing and RSVP-TE in production environments.
• To determine whether MPLS-TE outperforms IP-based TE in terms of path control, stability, and resource utilization.
• To provide empirical insights into the trade-offs between complexity, scalability, and reliability in TE solutions.

Proposed method

• The study evaluates constraint-based routing (CBR) and RSVP-TE for IP-based traffic engineering using operational network data.
• It compares these with MPLS-TE, which uses label-switched paths (LSPs) and explicit routing for traffic engineering.
• Performance metrics include path stability, setup success rate, and load balancing across links.
• The analysis is based on real operational network traces from a service provider, collected over multiple time periods.
• The authors use simulation and measurement-based analysis to compare routing convergence, path selection, and resource allocation efficiency.
• The study evaluates both static and dynamic TE configurations under varying traffic loads and network conditions.

Experimental results

Research questions

• RQ1 How do IP-based TE mechanisms like RSVP-TE compare to MPLS-TE in terms of path setup success and stability in real networks?
• RQ2 What are the scalability limitations of IP-based traffic engineering in large-scale operational deployments?
• RQ3 How does MPLS-TE ensure better load balancing and resource utilization compared to IP-based approaches?
• RQ4 To what extent do routing instabilities and suboptimal path selection affect IP-based TE performance?
• RQ5 What are the practical trade-offs between implementation complexity and control granularity in different TE architectures?

Key findings

• MPLS-TE demonstrated significantly higher path setup success rates and better load balancing compared to IP-based RSVP-TE.
• IP-based TE showed higher path instability and frequent rerouting under dynamic traffic conditions.
• MPLS-TE provided more predictable and deterministic path selection, crucial for service-level agreement (SLA) compliance.
• The study found that constraint-based routing in IP networks often failed to achieve optimal path selection due to limited global state visibility.
• MPLS-TE outperformed IP-based solutions in both setup time and resource utilization efficiency in operational environments.
• The authors observed that IP-based TE mechanisms were more prone to routing oscillations and suboptimal LSP establishment under high load.