The efficient management of complex distributed applications in the Cloud-Edge continuum, including their deployment on heterogeneous computing resources and run-time operations, presents significant challenges. Resource management solutions -- also called orchestrators -- play a pivotal role by automating and managing tasks such as resource discovery, optimisation, application deployment, and lifecycle management, whilst ensuring the desired system performance. This paper introduces Swarmchestrate, a decentralised, application-centric orchestration framework inspired by the self-organising principles of Swarms. Swarmchestrate addresses the end-to-end management of distributed applications, from submission to optimal resource allocation across cloud and edge providers, as well as dynamic reconfiguration. Our initial findings include the implementation of the application deployment phase within a Cloud-Edge simulation environment, demonstrating the potential of Swarmchestrate. The results offer valuable insight into the coordination of resource offerings between various providers and optimised resource allocation, providing a foundation for designing scalable and efficient infrastructures.

翻译：在云边连续体中高效管理复杂分布式应用（包括其在异构计算资源上的部署和运行时操作）面临重大挑战。资源管理解决方案——亦称为编排器——通过自动化和管理资源发现、优化、应用部署和生命周期管理等任务，同时确保期望的系统性能，发挥着关键作用。本文提出Swarmchestrate，一种受群体自组织原则启发的去中心化、应用中心编排框架。Swarmchestrate解决了分布式应用从提交到跨云边提供商的最优资源分配，以及动态重配置的端到端管理问题。我们的初步研究成果包括在云边仿真环境中实现了应用部署阶段，展示了Swarmchestrate的潜力。该结果为协调不同提供商间的资源供给与优化资源分配提供了重要见解，为设计可扩展且高效的基础设施奠定了基础。