The fourth industrial revolution advocates the reformulation of industrial processes to achieve the end-to-end (provider-customer) digitalisation of the industrial sector. As is well known, the industrial environment is very complex, where legacy systems must interoperate and integrate with modern devices and sensors. Communication among them requires specific and costly developments, so architectures based on data sharing and services implementation are considered one of the most flexible and appropriate technological solutions to gradually achieve the desired horizontal and vertical integration of the value chain. The design and deployment of data-intensive applications is not straightforward, therefore this paper proposes a model-based tool to characterise the different elements to be configured in an application and to make its deployment easier by generating configuration, orchestration and deployment files and sending them to the corresponding nodes for their execution. In few words, this article highlights the advantages of distributed and data-centric architectures to face the challenge of integration and interoperability in data-intensive complex systems and presents the extension of the RAI4 metamodel proposed in Martínez et al. (2021) that now allows specifying how, containerised or not, and where, on the cloud, fog, edge or on-premise, each service can be hosted according to its functional and non-functional requirements, mainly issues related with real-time, security and cyber physical hardware dependencies. For the sake of comprehension, a pseudo-real use case addressed to pre-process and store pollution data from environmental sensors installed in a smart city is described in detail, including different deployment settings.

中文翻译：

---

用于在混合环境中设计、配置和部署数据密集型应用程序的基于模型的工具：工业 4.0 案例研究


第四次工业革命提倡重新制定工业流程，以实现工业部门的端到端（提供商-客户）数字化。众所周知，工业环境非常复杂，遗留系统必须与现代设备和传感器进行互操作和集成。它们之间的通信需要特定且成本高昂的开发，因此基于数据共享和服务实现的架构被认为是逐步实现价值链所需的水平和垂直整合的最灵活、最合适的技术解决方案之一。数据密集型应用程序的设计和部署并不简单，因此本文提出了一种基于模型的工具来描述应用程序中要配置的不同元素，并通过生成配置、编排和部署文件并发送它们来简化其部署到相应的节点执行。简而言之，本文强调了分布式和以数据为中心的架构在面对数据密集型复杂系统中集成和互操作性挑战时的优势，并提出了 Martínez 等人提出的 RAI4 元模型的扩展。 (2021) 现在允许指定每个服务如何根据其功能和非功能需求（主要是与实时相关的问题）托管（无论是否容器化，以及在云、雾、边缘或本地的何处） 、安全性和网络物理硬件依赖性。为了便于理解，详细描述了一个伪真实用例，用于预处理和存储来自安装在智能城市中的环境传感器的污染数据，包括不同的部署设置。