NFV (Network Function Virtualization) promises to fundamentally change the nature of telecom networks, by completely separating the hardware from the software; network functions and services will be virtualized and will be delivered using standard IT hardware resources (processing, storage and bandwidth) from data centers in the cloud. This will be the real enabler for operators’ transformation into Digital Services providers, as they will be able to design and deliver new services much faster, configure digital experiences in real-time and slice their networks to deliver on-demand network capabilities corresponding to different applications (connected cars, Smart Cities, etc.) In brief this will allow CSPs to become the true enablers of the digital transformation of many industry verticals (transport, energy, smart cities, etc.) and to deliver new lucrative managed services to these verticals.
The deployment of NFV will be a long migration journey that will stretch over many years – probably a decade – during which operators will have to manage hybrid networks (virtualized and physical). NFV is still in standardization phase, with the key challenge being to define the architecture and interactions of management, orchestration and OSS functions that will enable the configuration, delivery and assurance of NFV infrastructure and end-to-end services. In particular the key question is how to ensure that automated NFV orchestration will always result in optimal network Quality of Service (QoS) and customer Quality of Experience (QoE). Some Operators have already started trials and proof of concepts, most of which revolve around virtualization of consumer set top box (CPE) or corporate services such as VPN.
The real, large scale and commercially meaningful adoption of NFV will become a reality only when proper management, orchestration and OSS architectures are agreed, implemented, deployed and validated. But this seems today a (very) long journey filled with challenges.
Among these challenges is lack of unified standard and the fragmented standardization work undertaken by multiple bodies including ETSI NFV, 3GPP, Broadband Forum, IETF, ITU-T SG 15, MEF, ONF, OPNFV and TM Forum. In particular the interface between MANO (Management and Orchestration) and OSS (Os-Ma-nfvo) is loosely defined and there is a debate over the nature of this interface and the respective roles of these systems. In fact recently a CTO of a major European Operator told me that he expected the standardization effort to take at least another two years before producing a meaningful result.
So taking a more pragmatic approach, let’s look at the key business requirements of NFV management and orchestration:
As stated earlier, one of NFV’s main benefits is to allow operators to dynamically configure and deliver digital experiences with various levels of QoE in response or in anticipation of certain demand patterns. This means that the NFV Service orchestration and management layer has to have comprehensive, continuous and real-time awareness of end-to-end network QoS and customer QoE and must have ability to address it in real time. In other words the NFV Service orchestration processes have to be commanded by the OSS Assurance and Analytics layer, using interface Os-Ma-nfvo.
In addition, the end-to-end QoS and QoE will be impacted by the performance of the different layers from service transactions to virtual functions (VNF) and NFV infrastructure (NFVi). And given the dynamic nature of NFV network topology and configuration, a tight integration between fulfilment (inventory) and Assurance processes/data with surgical and frequent configuration updates will be a must.
For the next 10 years networks will be hybrid, and so the management and orchestration layer cannot address the NFV part only. It has to manage and correlate both NFV and physical resources and services to achieve the desired end-to-end visibility and target digital experience.
In addition, it is likely that the virtual functions will be introduced gradually, by different vendors, and most likely with proprietary orchestrators. Also SDN controllers will be managing the underlying transport layer for dynamic resource control and allocation. So a governance orchestration layer will be required to resolve conflicts and maintain consistency and end-to-end visibility.
Now considering the above we believe that the right approach to fast track the deployment of end-to-end NFV services and unlock the potential of dynamically creating, managing and assuring digital experiences is to introduce such an “end-to-end Digital Experience Orchestration layer” that will bridge both NFV MANO and OSS.
This layer will notably embed real-time predictive analytics, dynamic experience/service/network graph models and advanced policy engine. It will determine the optimal service orchestration decisions aimed at maximizing the digital experience and the service quality while not compromising CSP’s business policies, and then automatically send the NF scaling decisions to the service and domain orchestrators via the OS-Ma-Nfvo interface, as well as send the physical or legacy EMS controlled network modifications decisions to the fulfilment and configuration layer via the 3GPP CM interface. In other words it will act as an intelligent decision support layer for service orchestration, with the aim of maximizing the digital experience at all times.
Full automation will enable real-time reactivity, massive complexity management and order of magnitude operational efficiency gains compared to today’s operations.
Its heavy reliance on predictive analytics will allow it to take prescriptive actions such as configuring the network in anticipation of certain digital experience demands (e.g. very low latency/low bandwidth services) at a given location at a given time.
It will operate in full closed loop, continuously monitoring and reporting the impact of the decisions made on the digital experience and QoS, and adjusting accordingly. This will provide the critical command and control capability that CSPs need to retain as they migrate to NFV.
It will be completely operated in private or public cloud, with Elastic computing and storage, Open and standard (REST) API implementations such as TMF and 3GPP interfaces allowing for distributed information collection and processing and lifecycle management.
This layer is designed for true end-to-end management of all the services and network technology domains and layers. It will smoothly integrate new VNF, SDN controllers and NFV orchestrators as they are deployed by the CSP. And it will give the flexibility to the CSP to execute their physical – virtual migration roadmap without pressure and continue operating their physical infrastructure for as long as they wish.
This concept was demonstrated in June 2015 as part of MYCOM OSI’s catalyst demo “Closing the Loop: Customer Experience, Policy and Virtualization” at the TM Forum show in Nice.