Automating the deployment of cloud-native applications over multiple platforms

Deployment automation is crucial nowadays, and a plethora of deployment automation technologies have recently been released. The Essential Deployment MetaModel (EDMM) and the ecosystem of its accompanying tools were proposed to ease the migration from one technology to another. EDMM allows to describe multi-component application deployments in a technology-agnostic way. The EDMM Modelling tool enables editing the specification of the deployment in a graphical environment. The EDMM Transformation Framework is capable of transforming the EDMM specification of a deployment into the specific files required by a chosen declarative deployment technology to actually enact the application deployment. A Decision Support System (DSS) is also available, to support developers in selecting the technologies that can actually support a specified deployment. Unfortunately, the ecosystem of tools around EDMM was not integrated, hence requiring developers to set up different environments for different tools, and to switch from a tool to another. In addition, the DSS only allowed to understand which deployment automation technologies supported the current deployment, without any further suggestion. The thesis presents two contributions tackling the aforementioned issues. We first discuss the design and development of a standalone solution integrating the ecosystem of tools accompanying EDMM. We then present an extension of the DSS. It now allows developers to automatically adapt the deployment of a multi-component application to enable its transformation to a technology-specific model, if the deployment automation technology was not already supporting the modelled deployment. We finally assessed the integration and the enhanced DSS by a case study comparing different deployments of a third-party multi-component application performed with and without the integrated EDMM ecosystem