NFV (Network Function Virtualization) is a transformative technology whose advent has been highly anticipated for 3 years now. NFV Management and Orchestration (NFV MANO), in turn, is the heart and brain of the NFV architecture. In this blog, I’ll start with a high level description of the NFV MANO architecture followed by challenges that the industry faces in adopting this architecture.
NFV, as the name suggests, is the virtualization of network functions. In other words, it decouples the software implementations of network functions from the underlying physical hardware resources so that they can be hosted on commodity servers or virtual machines. The obvious benefits are cost optimization, flexibility and accelerated time-to-market. By leveraging NFV, service providers don’t have to engineer their networks for peak traffic rates and can instead scale their network functions as required without making a long-term commitment to resources.
The virtualized software implementation of a network function is called a Virtual Network Function (VNF). Multiple VNFs can be hosted on a single VM, or the same VNF can be deployed on more than one VMs. Running a VNF across multiple VMs may be desirable for fault-tolerance, load sharing and scalability. Additional VMs can be deployed at runtime to “scale out” the VNF without disrupting the service, also known as “horizontal scaling”. “Scaling in” means the exact opposite i.e. tearing down one or more VMs at runtime. On the other hand, “scaling up” or “scaling down”, also known as “vertical scaling”, implies adding to or removing resources from one or more of the already deployed VMs.
NFV sets the stage for network equipment manufacturers and service providers to transform their offerings from traditional monolithic architectures to future-proof cost-optimized cloud-based solutions and micro-services. It has wide-spread applicability spanning across virtualized mobile core networks (e.g. vEPC), cloud RANs (e.g. eNodeB), Edge networks (e.g. vCPE), IOT solutions, smart homes, connected cars, and much more.
Since NFV proposes a fresh network view with new network elements and relationships, there exists a need for new management and orchestration functions. This is where NFV MANO comes into play.
NFV MANO, a working group of the ETSI, provides an architectural framework for management and orchestration of all resources in a cloud datacenter (compute, network and storage), virtual network functions (VNFs) and network services. The primary intent is to allow flexible and quick on-boarding of network functions and subsequently managing their lifecycle, i.e. instantiating, upgrading, scaling up/down, scaling out/in, and tearing down the network functions. Other advanced capabilities include auto-healing and live migration.
ETSI NFV MANO Architecture
The ETSI NFV framework consists of three major components:
- Network Functions Virtualization Infrastructure (NFVI)
- Hardware and software components including compute, storage and networking, and the virtualization layer (hypervisor).
- Virtual Network Functions (VNFs)
- Management and Orchestration (MANO)
NFV MANO component is further comprised of the following functional blocks:
- Virtualized Infrastructure Manager (VIM)
- Controls and manages NFVI compute, storage, and network resources
- OpenStack is a classic example of VIM
- NFV Orchestrator (NFVO)
- Manages end-to-end network service (NS) over multiple VNFs
- Orchestrates NFVI resources across multiple VIMs
- VNF Manager (VNFM)
- Manages lifecycle of one or more VNF instances
In addition, NFV MANO also maintains the following data repositories:
- NS Catalogue
- List of Network Service Descriptors (NSD), Virtual Link Descriptors (VLD) and VNF Forwarding Graph Descriptors (VNFFGD)
- VNF Catalogue
- Repository of VNF Descriptors (VNFD)
- VNFD is a deployment template that describes a VNF
- NFV Instances Repository
- List of VNF and NS records storing dynamic information and state of VNF/NS instances
- NFVI Resources Repository
- Tracks NFVI available/reserved/allocated resources against NS and VNF instances
Now that we have talked about the basic architecture of NFV MANO, let us discuss the challenges that have so far kept commercial deployments from becoming a reality, even though most service providers acknowledge the necessity of NFV and have long been committed to the transition.
All physical network functions are not going to be ready for virtualized deployments during the early phases of transformation to NFV. This makes it necessary for MANO based network management system to be able to manage a hybrid network containing both physical network functions (PNFs) and virtual network functions (VNFs) from multiple vendors. Now that’s not easy, especially because PNFs do not belong to the world of virtualization, let alone orchestration.
The NFV MANO architecture proposes multiple management nodes such as VNFM, VIM and NFVO in contrast to the traditional networks that only had an EMS or at most an additional OSS/BSS. As a matter of fact, MANO doesn’t eliminate the need for an EMS or OSS either, and expects all five of these components to work in coordination. For NFV MANO to penetrate into existing deployments, it would need to integrate with their OSS/BSS solutions and there, unfortunately, is no standard way to do that.
While ETSI has proposed a reference architecture and the VNF and NS descriptor template definitions, it has not yet standardized the VNFM, NFVO and the corresponding interfaces due to which many vendors have already built proprietary solutions. Some of them do not even comply with the ETSI defined descriptor templates. The interoperability issues this has caused is concerning for both MANO solution vendors and VNF vendors.
On the other hand, VNF vendors have learnt that just virtualizing a traditional solution does not enable it for NFV deployment. In most cases, the solution needs to be re-architected so as to leverage the benefits of NFV and be a part of a holistically orchestrated service. To make it worse, not all virtualized solutions can guarantee the five-9’s telco-grade availability and reliability required for performance sensitive time-critical network services (refer to one of my previous blogs here).
Both service providers and network equipment vendors are constantly reinventing their business models. With these moving variables, it is not easy for NFV MANO to prove its case in the practical world.
The challenges however do not preclude the successful adoption of NFV for as it makes businesses resilient due to its inherent flexibility and agility.
Altran has a rich experience of system integration in networking, communication and cloud domains. It has a long history of successfully delivering quality stop-gap solutions to ensure a smooth transition to a new technology. Altran can play a vital role in bridging the gap between NFV MANO solution providers, network equipment manufacturers and network service providers, and help integrate multi-vendor and open-source solutions.