Orchestration in Fog Computing: A Comprehensive Survey

Abstract

Fog computing is a paradigm that brings computational resources and services to the network edge in the vicinity of user devices, lowering latency and connecting with cloud computing resources. Unlike cloud computing, fog resources are based on constrained and heterogeneous nodes whose connectivity can be unstable. In this complex scenario, there is a need to define and implement orchestration processes to ensure that applications and services can be provided, considering the settled agreements. Although some publications have dealt with orchestration in fog computing, there are still some diverse definitions and functional intersection with other areas, such as resource management and monitoring. This article presents a systematic review of the literature with focus on orchestration in fog computing. A generic architecture of fog orchestration is presented, created from the consolidation of the analyzed proposals, bringing to light the essential functionalities addressed in the literature. This work also highlights the main challenges and open research questions.

Autonomy and Intelligence in the Computing Continuum: Challenges, Enablers, and Future Directions for Orchestration

“…Costa et al [29] systematically review fog orchestration literature and propose a generic architecture for fog orchestration. They define fog orchestration as "a management function responsible for service life cycle.…”

“…Synthesizing the taxonomies in a number of recent works (see e.g. [28,29,25,23,26], and those mentioned in Section II-B), in this article, we take a holistic approach to the resources in the computing continuum. These resources, as depicted in Fig.…”

“…In this article, again synthesizing the concepts presented in recent studies on orchestration [28,29,25,23,26], we holistically define orchestration as the management of resources in the computing continuum, from fundamental resources to workflow resources, as depicted in Fig. 2.…”

“…Moreover, orchestration may rely on three different types of control topologies, the first a centralized control topology, a decentralized one, or a fully distributed (e.g., peer-to-peer) one [29].…”

Positioning Fog Computing for Smart Manufacturing

“…Making decisions about where and when on a shared platform to place services and tasks is known as orchestration [13]. For example, Kubernetes [14] employs a topdown model, where nodes are controlled by a centralized entity that makes all the decisions on orchestration based on policies and constraints defined by developers.…”