PyCloudSim: Modernized Cloud Computing Simulation Framework with the Incorporation of SFC

Cloud computing has become an important technology for providing on-demand resources to customers. However, the traditional networking architecture used in cloud computing faces several challenges, such as inflexible and static network infrastructure, high latency, and lack of visibility and control. To overcome these challenges, Software-Defined Networking (SDN) has emerged as a promising technology that separates the control plane from the data plane and provides dynamic control over network traffic. By combining both technologies, Network Service (NS), which emerged through Network Function Virtualization (NFV) as Virtual Machine or Container, has become the favourable deployment method for infrastructures such as 5G core or LoraWan stack. However, conducting research in related areas is still challenging due to the cost and technical difficulties of implementing appropriate laboratory environments. Simulation tool such CloudSim or CloudSim Plus is still focused on cloud computing and lack incorporation of the concept of a Service Function Chain. Hence, in this paper, we present a discrete event simulation framework with enhanced cloud computing simulation, network simulation, and support for NS and SFC with realistic API call simulation.

I. Introduction

Cloud computing has emerged as a crucial technology for delivering on-demand resources to customers [1]. However, the conventional networking architecture employed in cloud computing faces numerous challenges, including a rigid and static network infrastructure, high latency, and limited visibility and control [2], [3]. To address these issues, Software-Defined Networking (SDN) has emerged as a promising solution, decoupling the control plane from the data plane and offering dynamic control over network traffic [4]. SDN allows centralized controllers to efficiently and reliably manage dynamic changes in network topology. By combining SDN with cloud computing, Network Services (NS) can be implemented as Virtual Machines or Containers using Network Function Virtualization (NFV) and can establish Service Function Chains (SFC) within the ESTI MANO framework. For instance, in the context of 5G Open RAN, a solution like Open5gs requires deploying seven core components across a single host or multiple hosts, each with specific configurations for over thirty network interfaces. Despite significant advancements in SDN and cloud computing over the past decade, conducting experiments on real infrastructure remains costly (requiring server stacks and network equipment) and time-consuming (deploying platforms such as OpenFlow, Kubernetes, etc.). As a result, researchers have turned to computer simulations to facilitate their experiments.