Concerto reconfiguration

Component-based coordination applied to the deployment and reconfiguration of distributed databases

Concerto Galera Ansible Grid'5000 Python
Galera experiment Concerto-D Grid'5000

Concerto is a formal, component-based model for coordinating the lifecycle of distributed software: each component is described by places, transitions, and ports, and the coordination engine derives a correct parallel execution of deployment and reconfiguration operations.

This work is the experimental groundwork behind CoAnsible: before extending Ansible with coordination logic, you need a reference point on what coordinated reconfiguration actually buys you.

🧩 What I worked on

  • Running Galera cluster deployment and reconfiguration scenarios on Grid’5000, comparing Concerto-driven coordination against plain imperative automation
  • Translating existing Ansible playbooks into Concerto assemblies, and building the tooling that extracts the dependency structure from a playbook to generate the corresponding component code
  • Measuring reconfiguration time and parallelism on multi-node deployments, across several Grid’5000 sites

🔍 Why it is interesting

Configuration management tools describe what the state of a machine should be, but they have no first-class notion of when an operation on one node may proceed with respect to another. Adding a coordination model on top exposes the real parallelism of a reconfiguration, and makes the synchronization points explicit instead of implicit in the order of tasks.