Mini-SFC: A Comprehensive Simulation Framework for Orchestration and Management of Service Function Chains

In the continuously evolving cloud computing and network environment, service function chain (SFC) plays a crucial role in implementing complex services in the network with its flexible deployment capabilities. To address the limitations of existing SFC simulation tools, this paper introduces Mini-SFC, a modular simulation framework that supports both numerical and container-based virtual simulations, while also supporting online dynamic topology adjustments. As an open-source platform emphasizing user-friendliness, Mini-SFC facilitates rapid algorithm verification and realistic service deployment validation. By simplifying module design and providing standardized solver interfaces, Mini-SFC significantly shortens the learning curve for researchers and enhances the flexibility and scalability required for advanced SFC management and optimization. For readers interested in exploring or utilizing Mini-SFC, more information is available on the official project page.

💡 Research Summary

The paper introduces Mini‑SFC, a comprehensive, open‑source simulation framework designed to address the shortcomings of existing Service Function Chain (SFC) simulation tools. Modern cloud‑network environments rely heavily on SFCs to compose complex services by chaining virtual network functions (VNFs). While several simulators exist—numerical tools such as Virne and PyCloudSim, and container‑based platforms like Mini‑nfv and vim‑emu—each suffers from limited functionality: difficulty of use, lack of dynamic topology support, or high configuration complexity. Mini‑SFC tackles these issues through four key innovations.

Dual‑Mode Simulation
Mini‑SFC offers both a numerical mode and a container‑based virtual mode. In numerical mode, the framework uses SimPy for discrete‑event simulation and NetworkX to represent the substrate network as a graph whose node and edge attributes store CPU, memory, and bandwidth capacities. VNFs are modeled as resource‑consuming entities that update these attributes, enabling rapid large‑scale experiments and early‑stage algorithm validation. In container mode, Mini‑SFC leverages Containernet, Docker, and Open vSwitch to instantiate each VNF as an isolated Docker container. Real service times, packet processing delays, and inter‑container networking overhead are faithfully reproduced, providing a near‑realistic testbed for final performance evaluation. The authors note that on a single 8‑core, 32 GB machine, running more than 30 containers introduces noticeable latency and jitter, suggesting a hybrid approach where numerical simulations are used for scale and container simulations for fidelity.

MANO‑Inspired Modular Architecture
The framework adopts the ETSI‑NFV MANO reference architecture (NFV Orchestrator, VNF Manager, Virtualized Infrastructure Manager, and User Equipment Manager) but abstracts away the heavyweight configuration of full MANO stacks such as OSM. Each component is implemented as a lightweight Python class with clearly defined interfaces. The central “Solver” module receives two standardized inputs whenever an SFC‑related event occurs: (1) a description of the event (VNF types, resource demands, QoS constraints) and (2) the current network state (available node/link resources, topology). The solver returns a structured mapping table that specifies VNF‑to‑node assignments, routing paths, and resource allocations. This table is directly consumed by the NFVO and VNFM, which then carry out deployment or migration without additional orchestration logic. Consequently, researchers need only implement their algorithmic core while adhering to the input/output schema, dramatically lowering the learning curve compared with platforms that require deep knowledge of OSM or other MANO workflows.

Dynamic Topology Support
A standout feature of Mini‑SFC is its ability to modify the network topology at runtime. An event‑driven mechanism triggers updates in the “Topo” module, which inherits from NetworkX and can add or remove nodes/links, adjust capacities, or simulate failures on the fly. The solver automatically receives the updated graph, enabling experiments that reflect the highly dynamic environments expected in 6G space‑air‑ground integrated networks, mobile edge computing, or large‑scale IoT deployments.

User‑Friendliness and Low‑Cost Deployment
Mini‑SFC is released on GitHub with extensive documentation, tutorial wikis, and example scripts. It can be installed on a single workstation, making it accessible to research groups with limited budgets. The modular design encourages contributions: new solvers, custom trace modules, or additional MANO components can be added without altering the core codebase.

The paper provides a comparative table (Table I) showing that Mini‑SFC uniquely supports both numerical and container simulations, dynamic topology, open‑source licensing, and ease of use—all simultaneously. Detailed descriptions of each module (NFVO, VNFM, VIM, UEM, Event, Net, Topo, Trace, Solver) illustrate how they interact in a typical workflow: load a scenario configuration, initialize the substrate and service topologies, schedule events, invoke the solver upon each event, and finally log performance metrics via the Trace module.

Two illustrative use cases are presented. The first demonstrates a pure numerical experiment where the substrate graph, service chains, and VNF templates are defined via Python dictionaries; the solver computes embeddings, and the VNFM updates the graph accordingly. The second showcases a container‑based run where Docker containers are instantiated for each VNF, traffic is generated, and real‑time measurements (latency, throughput) are collected.

In conclusion, Mini‑SFC bridges the gap between fast, abstract algorithm testing and realistic service deployment validation. By unifying dual simulation modes, providing a MANO‑aligned yet lightweight modular stack, and enabling online topology changes, it lowers barriers for SFC research and accelerates the transition from theory to practice. Future work outlined includes scaling to multi‑node clusters, parallel event processing for higher performance, and integration of AI‑driven optimization modules.