SDFog: A Software Defined Computing Architecture for QoS Aware Service Orchestration over Edge Devices

TL;DR

The proposed system, called "Software Defined Fog" (SDFog), abstracts connected entities as services and allows applications to orchestrate these services with end-to-end QoS requirements and aims at developing an integrated system to realize the software-defined control over fog infrastructure.

Abstract

Cloud computing revolutionized the information technology (IT) industry by offering dynamic and infinite scaling, on-demand resources and utility-oriented usage. However, recent changes in user traffic and requirements have exposed the shortcomings of cloud computing, particularly the inability to deliver real-time responses and handle massive surge in data volumes. Fog computing, that brings back partial computation load from the cloud to the edge devices, is envisioned to be the next big change in computing, and has the potential to address these challenges. Being a highly distributed, loosely coupled and still in the emerging phase, standardization, quality-of-service management and dynamic adaptability are the key challenges faced by fog computing research fraternity today. This article aims to address these issues by proposing a service-oriented middleware that leverages the convergence of cloud and fog computing along with software defined networking (SDN) and network function virtualization (NFV) to achieve the aforementioned goals. The proposed system, called "Software Defined Fog" (SDFog), abstracts connected entities as services and allows applications to orchestrate these services with end-to-end QoS requirements. A use-case showing the necessity of such a middleware has been presented to show the efficacy of the SDN-based QoS control over the Fog. This article aims at developing an integrated system to realize the software-defined control over fog infrastructure.

Figures (7)

• Figure 1: High level overview of Software Defined Fog environment. The SDFog Controller is logically centralized and may be implemented as a multi-domain controller
• Figure 2: Architecture of a fog node showing the proposed Middleware and Orchestration Engine. User applications and VNFs run on the same platform through hypervisor.
• Figure 3: Schematic representation of steps involved in task deployment following service discovery. Flow creation computes flow path based on current state of network topology. Flow installation sends VNF instantiation and flow table update commands to concerned fog nodes.
• Figure 4: Illustration of use-case Health Smart Home. Both Normal and Emergency modes of operation and the services involved in these modes are shown.
• Figure 5: Sequence diagram for Emergency operation mode in HSH use-case. Distributed service discovery through communication between instances on different fog nodes is shown.