Agents in the Computing Continuum: the MLSysOps Perspective.

Multi-agent systems (MASs) have gained considerable attention in the field of distributed computing due to their ability to provide technical interoperability, resource sharing and flexible coordination. Consequently, MAS are well-suited to address the challenges posed by the distributed and heterogeneous nodes within the device-edge-cloud continuum, including orchestration and standardization, optimal resource allocation,micro service placement policies, security and privacy.The objective of this study is to introduce the MLSysOps project, which aims at the autonomous management of the entire continuum tackling some of the challenges mentioned before. MLSysOps utilizes a hierarchical agent-based AI architecture to interface with the underlying resource management and application deployment/orchestration mechanisms. A comparative analysis is conducted between the existing related work and the proposed framework, highlighting the peculiarities and advantages of our approach.