Real-time system upgrade in cluster-based virtualized environments**

物理网络资源虚拟化在下一代网络的业务和网络架构设计中起着至关重要的作用。它可以根据最终用户的需求灵活高效地使用网络资源。它提供基础设施即服务（IaaS），以简化不同应用程序对许多物理网络资源的整合和共享。该基础设施依赖于高效的资源编排机制（如资源适配、系统升级和监控）来满足不同实体（消费者、服务提供者和网络基础设施提供者）的需求。本方案旨在解决基于集群的IaaS中系统实时升级的相关问题。主要目标是定义和验证新的方法（例如，模型、算法和方法），以支持分布式集群内无停机和无热点的实时主机管理程序升级。与现有的解决方案不同，这些方法将对各种动态虚拟化环境做出反应，在这些环境中，资源可以根据应用程序的需要动态调整（例如，资源扩展和迁移）。建议的研究活动将有助于1)降低实体基础设施的成本和提高服务的可用性；2)增加云软件升级的灵活性；3)通过提出的解决方案在实际IaaS平台的实施和部署，为开源云管理平台做出贡献。****

Physical network resource virtualization plays a crucial role in the design of service and network architectures for next generation networks. It enables flexible and efficient use of network resources according to end-user needs. It provides Infrastructure as a Service (IaaS) to ease the consolidation and the sharing of many physical network resources by different applications. This infrastructure relies on efficient resource orchestration mechanism (e.g., resource adaptation, system upgrade and monitoring) to meet the requirements of different entities (consumers, service providers and network infrastructure providers). This proposal aims to tackle the issue related to real-time system upgrade in cluster-based IaaS. The main objective is to define and validate novel approaches (e.g., models, algorithms and methodologies) to support real-time host hypervisors' upgrade within distributed clusters with no downtime and no hotspots. Unlike the existing solutions, these approaches will react to various dynamic virtualized environments where resources can be adapted dynamically (e.g., resource scaling and migration) according to applications' needs. The proposed research activities will 1) help reducing physical infrastructure cost and enhancing service availability; 2) add more flexibility for cloud software upgrade; 3) contribute to open source cloud management platform though the implementation and the deployment of the proposed solutions in real IaaS platform.****