This article presents the Wolphin joint R&D effort between French labs (LIP6, LIPN) and companies (Alter Way, Objectif Libre, Gandi) to tackle the management and scheduling optimization problems in an unknown context, yet critical due to their exponential usage, of containers based infrastructures. The project ran from Jan 2017 to Feb 2019. First, we analyze the differential factors from the legacy VM (Virtual Machine) virtualization and scheduling principles to their containers counterpart. Second, this article showcases research and industrial strategies to optimize resource management for elastic loads, with offline and online demands and model-based approaches. Last, we provide with insights through an Open Source prototype implementation and we synthesis the research papers we published during the project duration. Finally, the paper concludes with some perspectives for the future and discusses lessons learned from the joint effort. Overall, the paper is reporting the Wolphin experience and results, a project managed by industrialists. It serves as a crossroad between academic researchers and software professionals in cloud technologies.