In recent years, a new trend of advanced applications with huge demands in terms of Quality of Service (QoS) is gaining ground. Even though Cloud computing provides mature management facilities with ubiquitous capabilities, novel requirements and workloads, foisted by new services, start to expose its weaknesses. In this context, a new Information and Communication Technologies (ICT) trend aims at pushing computation from the Cloud to be much close as possible to data sources, raising in the evolution of new paradigms namely Fog and Mist computing. Specifically, the Fog computing paradigm exploits powerful nodes such as servers, routers, and cloudlets that are coupled with the end devices or their access networks accordingly; they are ”relatively” close by the data sources. Whereas Mist computing, which is a lightweight form of Fog computing, pushes the resources even closer. Precisely, Mist computing uses particular nodes that could reside within the same network (e.g., Local Area Network (LAN)) as the end-devices. Considering the advancement that the hardware is knowing nowadays, Fog and Mist nodes are seen suitable to provide resources such as processing, storage, and networking in the proximity of data sources; thereby, the requirements of the new services could be met. Together with the Cloud, the Fog and Mist paradigms introduce a stacked architecture for data processing where a data pre-processing could be performed at the Mist level, then offloaded vertically to the upper layers (i.e., Fog nodes or the Cloud). In these circumstances, it is fundamental to build a management system able to provision efficiently the Fog/Mist-based applications. For this purpose, the Operating System (OS)-level virtualization using containerization technologies, considering its light footprint, fits as a suitable solution to provide Fog/Mist services. The industrial-grade Cloud middlewares, such as OpenStack, which is a reference architecture for Infrastructure-as-a-Service solutions, are still far away from incorporating this new trend. This article proposes an OpenStack-based middleware platform through which containers can be deployed/managed at the Fog/Mist levels.

中文翻译：

---

基于云的网络边缘容器部署和迁移使能机制

近年来，对服务质量（QoS）提出巨大要求的高级应用新趋势正在兴起。尽管云计算提供了成熟的管理设施，功能无处不在，但在新服务的强加下，新的需求和工作负载开始暴露出它的弱点。在这种情况下，一种新的信息和通信技术 (ICT) 趋势旨在推动来自云的计算尽可能接近数据源，从而推动新范式的发展，即雾计算和雾计算。具体来说，雾计算范式利用了与终端设备或其接入网络相应耦合的强大节点，例如服务器、路由器和小云；它们与数据源“相对”接近。鉴于雾计算是一种轻量级形式的雾计算，将资源推得更近。准确地说，Mist 计算使用可以与终端设备位于同一网络（例如，局域网 (LAN)）内的特定节点。考虑到当今硬件的进步，Fog 和 Mist 节点被认为适合在数据源附近提供处理、存储和网络等资源；因此，可以满足新服务的要求。与云一起，Fog 和 Mist 范式引入了用于数据处理的堆叠架构，其中数据预处理可以在 Mist 级别执行，然后垂直卸载到上层（即，雾节点或云）。在这些情况下，构建一个能够有效配置基于雾/雾的应用程序的管理系统至关重要。以此目的，使用容器化技术的操作系统 (OS) 级虚拟化，考虑到其占用空间小，适合作为提供 Fog/Mist 服务的合适解决方案。工业级云中间件，例如作为基础设施即服务解决方案参考架构的 OpenStack，距离融入这一新趋势还有很长的路要走。本文提出了一个基于 OpenStack 的中间件平台，通过该平台可以在 Fog/Mist 级别部署/管理容器。