The Cloud Computing paradigm has brought a new way of building application systems by strongly decoupling applications and the platforms on which they operate. This is particularly thanks to virtualization techniques (virtual machines, containers, and soon WASM) and the associated orchestration technologies (OpenStack, Kubernetes, and CloudWASM), which make it possible to port, deploy, and manage applications on heterogeneous resources and at large scale.
The objective of this thesis is to provide strategies and mechanisms to ensure the reliability of virtualized systems. In particular, orchestrating these systems involves managing multiple tasks such as deploying containers, scaling resources, managing updates, handling errors, and balancing the load across different nodes. Therefore, the orchestrator(s) play a critical role in the successful implementation of virtualized processing.
In this context, we aim to build a modeling and analysis/remediation engine that ensures SLOs in container orchestration both at design time and runtime. The ultimate goal is to enrich the orchestration process with a system that guarantees and optimizes the management of virtualized resources.