Virtual deployment of virtual networks in Network as a Service

Abstract

Listen

Translate article

While telecommunication and networking are driven by massive adoption of virtualization and the emergence of network programmability, the process of deployment remains an important phase in virtual networks (VNs) life-cycle. We raise through this chapter the question about the importance and opportuneness of redesigning VN deployment process in this new ecosystem. For that, we have first raised the impacts and advantages brought by Cloud Computing, SoftwareDefined Networks (SDNs) and Network Function Virtualization (NFV) over network architectures and operations. We have then focused on the notion of Network-as-aService (NaaS) which integrates these new technologies. We have come to define the role and the characteristics of NaaS. This way, we have been led to distinguish two issues for the implementation of VNs. The first one is the actual process of deployment of VNs at a network service delivery level that takes into account the applicative flows, and the second one is the placement process at the physical network infrastructure. Our objective is to present the “Virtual Deployment” as a phase within the VN deployment process. We define the virtual deployment phase using NaaS architecture. Our virtual deployment proposition takes into account the properties of flexibility, adaptability and dynamicity that are vital for the NaaS, where Cloud, SDN and NFV represent major building components in operators' network architectures. Thus, the virtual deployment of a VN is meant to integrate a response to a Service Level Agreement (SLA) request. Indeed, it supports the adaptation of a VN to integrate, starting from the deployment phase, network-level Quality of Service (QoS) constraints as a response to the service level agreement (SLA) request. To automate the virtual deployment, we propose to introduce the QoS constraints at the design phase. Therefore, through a Network Application Programming Interface (API), a network orchestrator will be able to consider “On Demand Services” and “User-Controlled services.” Also, we rely on the abstraction of the Network Operating System (NOS) to address the need for an ability to choose and adopt the most efficient network control functions. Thus a personalized VN is consolidated, which takes into account all network constraints coming like requests from SLA and from usage. We end this chapter by presenting technical tools that can be used for this purpose but they still have some limitations. The conclusion highlights the strengths of our proposal and introduces the perspectives.