Towards Cognitive Microservice Orchestration in the Multi-Cloud Continuum

Abstract

The proliferation of decentralized applications calls for complex deployments across the highly heterogeneous Cloud Continuum, an overlay infrastructure encompassing traditional public/private clouds, fog clusters, and edge devices. Orchestrating distributed applications within such a non-uniform environment presents significant challenges, particularly in ensuring the fulfillment of critical Quality of Service (QoS) requirements. Achieving the objectives posed by this research challenge requires addressing several key needs, including the development of a robust system for virtualizing heterogeneous resources spanning multiple domains, the design of an intelligent scheduler capable of implementing deployment schemes that respect application-specific QoS requirements, and the provision of an orchestration mechanism that simplifies the task of specifying application functional and non-functional requirements for the application owners while automating the application roll-out. In this work we propose a framework that integrates existing tools that individually address the aforementioned needs by blending them in a unified manner. Preliminary experiments conducted on a software prototype demonstrate the feasibility of the proposed approach.