14:00
17:00

Today, enterprises are adopting the Network Functions Virtualization (NFV) paradigm en masse to deploy, maintain, and scale their network services in an automated, dynamic, and flexible manner.
NFV advocates implementing physical network functions such as routers and firewalls as software instances that can run on standard servers.
With this approach, many companies are moving their networking functions out of their server rooms and into the cloud. In this way, they reap the enormous benefits of cloud computing, which include high availability, flexibility and resiliency.
However, the lack of trust in the cloud by enterprises inhibits the adoption of outsourcing network services to the cloud. For example, a rogue cloud provider may deploy an enterprise's network service with less than the required amount of resources, and obscure this violation due to the opacity of the cloud. In addition, because the cloud software stack is complex in nature, the risk of configuration errors occurring on network services remains significant. As a result, organizations that outsource their network services to the cloud lack assurance that their network services will conform to their specifications. Thus, widespread adoption of network services outsourcing to the cloud requires first and foremost solving this trust issue between enterprises and the cloud.

This thesis supports the idea that network service verification remains the relevant answer to increase enterprises' trust in the cloud. The objective of this thesis is to study potential service anomalies in NFV and propose verification mechanisms to detect these anomalies. First, this thesis presents a taxonomy of network service anomalies in NFV as well as a review of their verification methods. Second, we present VeriNeS, a verification system that observes and analyzes the behavior of the NFV Orchestrator to detect network service anomalies.

Amphi LaBRI