In order to thrive in an increasingly competitive environment, service pro­viders need to offer their customers a dynamic catalogue of competitive new services that can be deployed and modified in hours rather than months, and can support their growing expectations from user-defined networks. It is, thus, fundamental for service providers to migrate to an internet protocol (IP) network infrastructure and a commercial off-the-shelf hardware to eliminate vendor lock-in for achie­vi­ng these goals. An additional requirement is to bridge the divide between business and network systems for ensuring an end-to-end quality customer experience. In this regard, network functions virtualisation (NFV) technology can help operators to transition from hardware to software-defined platforms, resulting in increased automation.

Need for NFV

Existing networks, which contain a variety of proprietary appliances from numerous vendors, create a barrier to new service introductions, drive up the cost of meeting capacity demands and are difficult to maintain. These obstacles negatively impact service providers’ ability to be effective in a competitive landscape that consists of both existing and over-the-top new service providers. The traditional network architecture poses the following challenges.

  • Difficulty in launching new services: The introduction of additional appliances in the network, often provisioned with significant over capacity, substantial integration efforts, shipment to numerous physical sites, manual installations and end-to end testing, is required for launching a new service. These activities being time-consuming and entailing large upfront investments are difficult to market and offer lower returns on investment.
  • High costs: The reliance on static appliances raises the cost of meeting capacity demands and specific service-level agreements. Further, the lack of network programmability and automation makes real-time capacity management and resource pooling impossible. Service providers need to anticipate the demand in advance and roll out networks designed to handle peak usage and unexpected traffic bursts.
  • Complexities involved in changing and maintaining ongoing services: Service providers incur high operational costs that include a dedicated spare parts inventory per appliance, technical support skills, shipping costs, larger floor space and increased power consumption due to the need to support numerous proprietary appliances across sites. Moreover, it is inherently difficult and expensive to maintain the service fabric across different devices in the network.

The above-mentioned challenges and the increasing need to cater to the accelerating capacity demands are driving a paradigm shift towards NFV. The technology is expected to remove new service introduction barriers and directly connect business systems with the network, enabling managers to deliver tailored customer experiences quickly. NFV will also help in building user-defined network services to provide customers with more control over their experience, more service variety and choice, flexible pricing models and differentiated quality of service. In addition, it will help reduce network operations and service maintenance costs.

Key challenges in NFV deployment

Successful navigation to NFV will be challenging for service providers. The path to NFV requires a shift of technology, operational and organisational process shifts, and new business models. These challen­ges can be broadly classified as:

  • Technological challenges: NFV-based systems will initially need to co-exist with legacy physical systems as the migration will entail a lengthy transformation pro­cess that has to be done gradually to minimise risk. During this phase, service providers will have to manage hybrid networks, under which service orchestration is exceedingly complex.

Another key technological challenge is the lack of standardisation. Many network function providers use proprietary systems to manage their virtual network functions. This creates difficulties in identifying the open environment that is required. It is, therefore, necessary to select the right vendor and bring on board new service functions for service integration, chaining and management.  Service providers will need to carefully select vendors that are capable of helping them integrate many software-based systems into a flexible NFV services fabric.

In addition, while introducing new virtual use cases one at a time, service providers risk retaining the current state of multiple silo networks that require numerous orchestration and management systems. This minimises the opportunity to reduce complexity and take advantage of full-cloud elasticity and agility benefits. To mitigate this risk through advanced planning and vendor selection, the transformation needs to be done in a thought-through way.

  • Operational and organisational challenges: The skill sets for running a hybrid, virtualised and physical network will have to include cloud-based capabilities such as OpenStack, software lifecycle management and a modified approa­ch to the network operations centre. Moreover, there will be a need to simplify and re-engineer logistical barriers for new services. The general concept of lifecycle service delivery will also have to be redesigned. Further, new employees will be required to create and launch new services and new methods for testing service success in some situations. While implementing the operational, organisational and procedural changes, service providers are likely to encounter challenges related to employee retention and organisational stability.
  • New business model challenges: To leverage the cost efficiencies of the new flexible virtual network, service provi­ders will need to change the way they work with their suppliers to a “pay-as-you-go” or “pay-as-you-grow” model. Me­chanisms and models have to be put in place to accurately reconcile with vendors as network changes result in the use of one vendor’s virtual network function over another. New charging models, such as per virtual machine, per CPU or per CPU socket, would need to be established. Similarly, revenue-share models, in which the network equipment providers are no longer providers but partners in the ecosystem, will have to be introduced.

Besides, new pricing models will result in software licence management challenges. To this end, service providers will need a method and a mechanism to track licences and provide the ability to attach licences to the virtual machines used. Currently, there is a one-to-one ratio, one licence per one proprietary NEP, but this model will no longer be viable.

As the network opens up, service providers will need to manage a growing number of suppliers and multiple vendors can be used for each function. They will also need to continuously differentiate between suppliers on a cost performance basis to provide the best services at the lowest possible cost.

Further, justification of the transformation business case will be a challenge. Service providers will need additional visibility in their current network spend and the ability to compare it to the cost of a new, never-tried-before architectural model in order to measure the potential cost savings. The required management and virtual infrastructure system investment may potentially offset the hardware savings. A strategic part of the business case for NFV will be based on new services and shorter time to market. Since the return on investment on new services is often difficult to calculate, it would be difficult to ascertain the adoption rate, that is, how much customers will be willing to pay for NFV services.

Conclusion

There is undoubtedly a business need for the NFV technology in the industry today. This can be attributed to the dynamic set of problems faced by service providers. However, looking at the key requirements for NFV and the technological, operational, organisational and business model challenges that service providers would face while shifting to this technology, it still remains unclear whether all telecom operators would be able to adopt this technology in the next two to three years.

Based on a white paper, “Bringing NFV to Life”, by Amdocs