Specifying and Placing Chains of Virtual Network Functions

Abstract: Network appliances perform different functions on network flows and constitute an important part of an operator's network. Normally, a set of chained network functions process network flows. Following the trend of virtualization of networks, virtualization of the network functions has also become a topic of interest. We define a model for formalizing the chaining of network functions using a context-free language. We process deployment requests and construct virtual network function graphs that can be mapped to the network. We describe the mapping as a Mixed Integer Quadratically Constrained Program (MIQCP) for finding the placement of the network functions and chaining them together considering the limited network resources and requirements of the functions. We have performed a Pareto set analysis to investigate the possible trade-offs between different optimization objectives.

• The paper presents a novel context-free language to specify VNF chains, accommodating complex orders and parallel modules.
• It uses an MIQCP formulation to optimize VNF placement by balancing latency and remaining data rate across network nodes.
• A heuristic approach significantly cuts computational time, making the deployment of VNFs in dynamic networks more practical.

Specifying and Placing Chains of Virtual Network Functions

The paper addresses the formalization and placement of Virtual Network Functions (VNFs) in network service chaining, presenting a methodological approach to network function virtualization (NFV). The authors propose a robust model utilizing a context-free language to specify complex chaining requests for VNFs, tailored to the demands of modern dynamic networks.

Key Contributions

1. Chaining Model and Specification Language: The paper introduces a context-free language for articulating VNF chaining requests, adeptly accommodating complex scenarios such as optional orders and parallel modules within chains. This formal approach supports the dynamic and flexible specification of ordered sequences and parallel processing requirements necessary for advanced network functions.
2. Placement Optimization via MIQCP: VNFs are placed on network nodes using a Mixed Integer Quadratically Constrained Program (MIQCP). This approach strategically maps request demands to available resources while optimizing for multiple objectives, such as maximizing remaining data rate and minimizing latency. The MIQCP formulation accommodates the constraints and interdependencies of VNFs, demonstrating a nuanced balance between computational resources and network characteristics.
3. Pareto Set Analysis and Heuristic Application: The paper evaluates potential trade-offs between different optimization goals. A heuristic is proposed to reduce computational complexity by prioritizing chaining configurations that minimize data rate requirements. Although this approach may slightly deviate from the absolute optimal solution, it significantly reduces computational time, showcasing a pragmatic approach to deployment.

Numerical and Theoretical Insights

• The heuristic method demonstrated a substantial reduction in computational time (e.g., optimizing a scenario within 13 minutes compared to 31 hours required for exhaustive evaluation), highlighting the practical applicability of the proposed model.
• Pareto analysis reveals the inter-objective trade-offs, emphasizing that while some placements optimize all metrics, others necessitate balancing between latency, data rate, and node utilization.

Implications and Future Directions

The formal model for VNF chaining and placement presented has significant implications for the deployment and management of VNFs in operator networks. By addressing flexibility in chaining and resource allocation, the work supports NFV adoption in diverse environments, such as broadband, mobile core, and data center networks.

• Practical Implications: Network operators could leverage this model to enhance the efficiency and responsiveness of network service deployments, potentially leading to lower operational costs and improved service scalability.
• Theoretical Implications: The methodology provides a structured template for further research into optimization models for NFV, particularly in exploring alternative heuristic strategies and optimization objectives.

Future research may focus on extending the model to encompass dynamic network conditions and further integrate with emerging paradigms such as Software-Defined Networking (SDN) to enhance programmability. Additionally, exploring distributed algorithms that exploit the modularity of VNFs could further optimize placement decisions in real-time scenarios.

Overall, this paper offers a comprehensive framework for addressing the challenges of VNF specification and placement, bridging the gap between theoretical formalization and practical applications in network function management.