CloudSimSFC: Simulating Service Function chains in Multi-Domain Service Networks

Service Function Chain (SFC) is widely adopted in Multi-domain Service Networks (MDSN) to enforce network policies on customer traffic. Great effort has been devoted to the research of SFC deployment strategies. In this context, simulators are helpful to the design and evaluation of those strategies. A good SFC simulator should be accurate, comprehensive and user-friendly for SFC simulation. However, existing simulators fail to satisfy the requirement due to two major drawbacks: First, they overlook the resource heterogeneity and performance instability of the MDSN environment. Second, their performance models for service functions are too simple to contain the important features such as traffic changing effect and packet queue. To overcome these drawbacks, we propose CloudSimSFC — a new SFC simulator that (1) simulates the performance fluctuations and server failure/recovery events in MDSN environment to align with the performance instability of real-world systems, (2) elaborates the modeling of service functions by incorporating the computation components such as CPU and queue, and the traffic changing effect which happens during packet processing, (3) employs scenario abstraction to simplify the definition of new (heterogeneous) simulation scenarios and supports standard service metrics like request–response time. With these features, CloudSimSFC can be used to simulate SFC run-time performance and evaluate deployment strategies. We introduce the system architecture of CloudSimSFC to explain the simulation principle. We conduct extensive experiments to evaluate the simulation accuracy of CloudSimSFC by comparing it with an SFC prototype. Experimental results confirm that CloudSimSFC achieves 95%+ accuracy in all the evaluation scenarios. We also evaluate the simulation running time and analyze the overhead caused by the proposed simulation features. Finally, we demonstrate the primary usage of CloudSimSFC with case studies.