Towards Orchestration in the Cloud-Fog Continuum

The growth of the Internet-of-Things has led to a rise in the need of computing power, storage, and network resources. As more data are being generated at the edge of the networks, the cloud model that enabled the affordable, on-demand, lease of these resources is ill-fitted to handle the volume and variety of data traveling to the core of the cloud and back. Some applications further showcase the limitations of the cloud by requiring strict low-latency communication and location awareness. Fog computing has been proposed as a solution to these issues that stem from the cloud's centralization. The fog is an emerging computing paradigm, conceived as an extension to the cloud, that aims to facilitate the creation of scalable infrastructures in the vicinity of the end-user. By decentralizing resources, it promises to optimize bandwidth consumption at the core and edge of the network while reducing latency between the service and the end-user. In this paper, we identify the requirements needed to orchestrate loads in the Cloud-Fog continuum and propose an architecture, built on available, open-source, components, that orchestrates loads with consideration to their geographical needs. We provide several levels of features (DNS-like service discovery, service mesh, health checks, encryption-as-a-service, among others) available to the operator and evaluate their quality-of-service implications, with respect to network latency and bandwidth, when compared to a simple deployment baseline.

Automated summary

The growth of the Internet-of-Things has led to increased demand for computing, storage, and network resources, surpassing the capabilities of the cloud model. Fog computing has emerged as a solution to optimize resource utilization and reduce latency, by decentralizing resources and enabling efficient data communication and processing closer to the end-user.