Fog computing is a geo-distributed model of computing in which local computing and storage infrastructures serve as intermediaries to better exploit data flows between connected objects (IoT) and a remote, centralized Cloud.
The SeMaFoR project - for Self-Management of Fog Resources - aims to offer a generic solution that can automatically and intelligently orchestrate the computing and storage resources of the Fog, with immediate applications to smart cities, smart factories, smart mobility, etc. Project stakeholders met for a first physical meeting on 30 and 31 August at IMT Atlantique (Nantes campus) to discuss the project, which will be launched remotely in March 2021.
Introduced in 2014 by Cisco, an American IT company originally specializing in network hardware and then servers, Fog Computing is a new paradigm aimed at decentralizing the Cloud by geographically distributing computing, storage and network resources, as well as the associated services. "Today, with the spectacular rise of the IoT and the deluge of data generated, the concept of geo-location offers many benefits: reduced latency, elimination of network bottlenecks, reduced bandwidth, better management of disconnections, control of data storage, etc.," explains Thomas Ledoux, professor of computer science at IMT Atlantique, who is leading the project. "Moreover, it is an evolution that responds to current environmental issues by encouraging local exchanges and reinforcing digital sobriety."
However, moving from the Cloud to Fog computing is not a weather phenomenon! Fog computing is a large-scale distributed system made up of different types of computing resources (e.g. sensors generating data, calculation/storage services, routers, etc.) ranging from your internet router (or even your smartphone) to data centers in California. But how should such a huge, dynamic, unreliable, heterogeneous system be managed?
A generic solution for the smart city, the smart factory, smart mobility...
The objective of the ANR SeMaFoR (Self-Management of Fog Resources) project is to model, design and develop a generic and decentralized solution for the self-administration of Fog resources. Due to its size, the management of such a system by a human administrator is impossible: Fog computing must be able to adapt to configure, optimize, repair, etc. by itself! Hence this intrinsic property of self-administration. Moreover, by thinking abstractly about the orchestration problem, the SeMaFoR project aims to offer a generic solution that could be suitable for any application domain, such as the smart city, the smart factory, smart mobility, etc., which are some of the target domains of Fog computing. "We expect the results to be impressive and wide-ranging," explains Thomas Ledoux. "For example, new distributed consensus algorithms, automatic generation and coordination of reconfiguration plans, an open source framework linked to market solutions (such as Kubernetes)."
The key players of the SeMaFoR project are LS2N - IMT Atlantique with its Stack, NaoMod and Tasc teams, LIP6 - Sorbonne University with its Delys team, and a mid-sized company with R&D expertise in Cloud and DevOps called Alter Way.
The project will last 42 months and receive €507,000 in ANR (French national research agency) funding.
by Pierre-Hervé VAILLANT
