The DevOps paradigm is a keystone of nowadays software development, and more and more companies are jumping in the DevOps train (as we also did with our preliminary work on continuous model-driven engineering). However, it currently lacks of properties such as reasoning, validation and automation that are classically addressed by the Modelling community.
We believe that several challenges faced by DevOps engineers can be tackled by modelling approaches, as well as the MDE community can benefit from including DevOps-like methodologies in its best practices. That’s why we plan to have a workshop that attracts industrial and academic partners, as a discussion forum where both communities can exchange and discuss these very topics: the very first Devops@Models workshop!
There are many ways you can participate in the event to helps us explore the following fundamental questions: Can modeling and MDE contribute to DevOps, or is modeling fundamentally in contradiction with the goals of agile and DevOps? And, conversely, given the important role of model-driven engineering in the conception and development of complex systems, can DevOps help in the continuous integration and evolution of modeling artifacts themselves?
In this sense, we are interested in the different aspects of DevOps and modelling, more specifically the relationship that exists between modelling and DevOps, and the opposite, between DevOps and modelling. In addition to research contributions, we are also interested in contributions related to the teaching of such topics. The workshop has the ambition to cover the following questions:
- Modelling for DevOps:
- Modelling can easily be linked to «dev», what is its relationship to «ops»?
- What aspects of DevOps can modeling contribute to?
- What is the role of modelling in DevOps?
- What concrete solution (e.g., method, process, tool) can modelling brings to DevOps?
- What kind of analysis or reasoning techniques are needed by the DevOps approach?
- How can modelling help to improve DevOps best practices?
- DevOps for Modelling:
- How to monitor a model-driven software in an «ops» context?
- How can DevOps help in achieving a kind of continuous model-driven engineering process?
- Are classical MDE tools and techniques ready for DevOps? Can we combine these tools with existing ones?
- How can DevOps be used to ease the adoption of MDE tools and techniques?
- What are the requirements for a model-driven tool to be DevOps compatible?
- What would be the benefits of considering model-driven development in a DevOps way?
FNR Pearl Chair. Head of the Software Engineering RDI Unit at LIST. Affiliate Professor at University of Luxembourg. More about me.
Modelling for DevOps:
Given the high degree of automation and declarative approaches in devops (container, composition, landscape setup, piping etc.) by all means this already can be considered “modelling”, and it’s in our graps to benefit from a unification of “lessons learned” from both worlds.
DevOps for Modelling:
In a way, modelling can be considered a straightforward logical continuation of “continous …” approaches. Actually, given the long history of modelling in this area, it is surprising that everything is automated but the development step. By chance, for the next miotope posting I’ve prepared a diagram, which tries to illustrate these thoughts: http://miotope.org/clarification.svg
Looking forward to models19!