DevConf.CZ 2025

Dean kelly

I am a software engineer from Red Hat's Emerging Technologies.
It is lovely to meet you!


Company or affiliation

Red Hat

Job title

Associate Software Engineer


Session

06-13
12:55
15min
CODECO: AI-Driven Orchestration for Multi-Cluster Edge Deployment
José Castillo Lema, Dean kelly, Alka Nixon

With the development of next generation IoT services, in particular services such as mobile IoT, or an Internet of Nanothings (IoNT), novel environments must be researched to combat the demands of this ever growing landscape. These new environments are expected to be highly mobile, involving small cell management, and large-scale deployments. Edge computing, and Decentralised Edge Cloud computing architectures play a key role in the Edge-Cloud continuum, as there is also an increase in service decentralisation.

CODECO, an EU funded Open Source Research project, is a Kubernetes plug-in aimed at optimizing application deployment onto edge devices through cognitive and cross-layer orchestration. By leveraging AI-driven decision-making, CODECO significantly improves deployment efficiency across the edge. This is achieved through Automated Configuration Management (ACM) coupled with Open Cluster Manager (OCM), which utilizes AI-generated recommendations from the Privacy-preserving Decentralised Learning and Context-awareness (PDLC) component. These recommendations are based on real-time resource metrics collected from available edge clusters, guiding the optimal deployment of applications to the most suitable edge cluster. Additionally, ACM offers a user-friendly interface, allowing users to easily deploy, monitor, and manage their applications.

Our dedication to Innovation and Research Community Engagement Programme encourages collaboration among developers, SMEs, and research communities. This talk targets stakeholders keen on advancing both AI Edge-Cloud orchestration. Our attendees will have the opportunity to understand CODECO's principles, objectives, key research contributions, open-source toolkits, AI prediction mechanisms, and training resources through a use case focused on Decentralised, wireless AGV Control for Flexible Factories including a demonstration of how an application is deployed at the optimal performance locations. Finally, we’ll discuss how the project and community can identify future improvements and contribute to the end goal of redefining the Edge cloud continuum.

Cloud, Hybrid Cloud, and Hyperscale Infrastructure
A113 (capacity 64)