I have taken reference from presentation at OPNFV summit, Beijing (China) – Distributed vnf management architecture and use-cases
There are 4 types of models to deploy NFV in any data center or telecom network service provider
- Centralized model
- Decentralized model
- Distributed model
- Distributed model (Edge NFV)
This is a basic and common model of NFV deployment for any service provider. In this model, all the virtualization functionality is located at service providers data center or operation premises and due to this fact, VNFs can be deployed at providers data center. An enthernet connection can be used to access the VNF from centralized server.
As you can see from image, all virtualization functionality taken place at customer premises and there are no VNFs are present at central data center.
In this model, virtualization functionality is distributed between data center and customer premises. VNFs can be deployed based on optimal feasibility, performance, reliability, scalability and cost considerations. VNF are randomly ordered, configured and chained.
Distributed model (Edge NFV)
This model integrates cloud edge into the NFV infrastructure. Virtualization functionality is distributed between customer premises, data center and cloud edge.