Thank you!

The VELVET days were a great success, and a superb opportunity for scientific exchange around deployment, reconfiguration, adaptation and DevOps. We were able to discuss various topics such as the decentralization of DevOps operations, automatic configuration, control theory applied to distributed software adaptation, formalization and more.


Special thanks to

  • our guest speaker Daniel Sokolowski from Switzerland,
  • Jolan and Eloi who helped me a lot with the organization,
  • all our speakers for their great talks and tutorials.


The days are free of charge (cocktail, lunch and breaks included), but you’ll have to finance the missions yourself. We have to limit the number of registrations for the days for financial reasons. The registration is closed.

These days are organized within the GDR GPL and the GT YODA (which I co-chair with Simon Bliudze and Rabéa Ameur-Boulifa). The ANR project SeMaFoR (led by Thomas Ledoux) also helps me finance these days.

The VELVET days will be held at IMT Atlantique, 4 rue Alfred Kastler, 44307 Nantes

A visioconference will be set up during these days, here is the link to connect:

Invited speaker

Daniel Sokolowski (St. Gallen Switzerland)

Daniel Sokolowski is interested in distributed systems and cloud technology from a Programming Languages and Software Engineering perspective. Currently, he is mainly focussing on advancing Infrastructure as Code (IaC), aiming at robust deployments in modern DevOps organizations.


Reliable Infrastructure as Code for Decentralized Organizations


In DevOps, organizational structures foster cross-functional teams that independently develop and operate applications. Yet, these applications often rely on each other, necessitating coordination between teams. Further, DevOps extends to the technical domain, where DevOps promotes automation throughout the software lifecycle. Infrastructure as Code (IaC) is key to this automation, applying software development practices to operational tasks like software deployment and configuration. Ensuring the reliability of both inter-team coordination and IaC programs is critical. Failures in coordination or errors in IaC can lead to system malfunctions or expose them to significant security vulnerabilities.

In this talk, I explore three aspects of achieving reliable IaC within decentralized organizations. Firstly, I address the challenge of automating deployment coordination in a decentralized environment. Existing approaches are typically manual or centralized. To overcome this issue, I present µs, automating coordination of IaC programs in a decentralized fashion. Secondly, I delve into the intricacies of safely updating components in distributed systems, emphasizing the need for coordinated timing to ensure distributed transactions are successful. I will discuss methods for implementing safe dynamic software updating (safe DSU) in a decentralized context and its integration into IaC. Thirdly, the correctness of IaC programs is crucial. I present Automated Configuration Testing (ACT), enabling efficient unit testing of IaC programs, its implementation ProTI, and how it is the stepping stone to verifying IaC programs.


13th of December (IMT Atlantique, J142)

14th of December (IMT Atlantique, J142)


As for the hotel, we’d advise you to stay in a hotel in the center of Nantes, which will be more convenient in any case and will be nicer for you.