CTO office Technology Briefing: Network Function Virtualization (NFV)
Network Function Virtualization is an emerging architecture for telecom networks that promises to bring the benefits of standardised open environments to Telecom Network Operators (TNOs). However change in this landscape is not easy so the question becomes is NFV a white night for the Telecom Network Operators or is it destined to be a white elephant?
By Nick Stavrakos and Chiradeep Vittal
Before the introduction of the PC, software ran on dedicated hardware built by companies like HP, IBM and Sun. Computers were expensive and lock-in was the order of the day. Then a white knight, in the form of the PC, came along to enable choice and drive down cost, through a standardized, open architecture. Today, the telecom networking world (think Internet) is in a similar situation where networks are built from specialized hardware that is expensive and inflexible.
Network Function Virtualization (NFV) is an emerging architecture for telecom networks that promises to bring the benefits of standardized open environments to Telecom Network Operators (TNO) who are struggling to maintain a viable business model as Over the Top (OTT) content providers fuel an ever increasing demand for data by consumers.
However change in the TNO landscape is not easy – so the question becomes, is NFV a white night for the TNO’s or is it destined to be a white elephant?
A Business Model At Breaking Point
TNO’s like AT&T, Verizon, BT, Telstra, France Telecom and NTT build and run the networks that are at the core of the Internet. This is a big business, valued in excess of US$2 trillion in 2012 by global analyst firm, Ovum. However the TNO's business model has been under continual attack as they transition from fixed line telephony to the data-centric, mobile Internet.
Today, TNO revenue is in large part derived by charging users for every bit of data that transits the network. Revenue growth is achieved through getting customers to consume more and more data. This model is a carry-over from the days of fixed-line telephony, where users were charged per local voice call or per minute on long-distance.
You could be forgiven for thinking that customers' ever increasing demand for data, would place the TNOs in a healthy business position. Unfortunately this is not the case. Due to the exponential rise in demand for digital media traffic, the growth in their costs is outpacing revenue growth, meaning that without a change in business model, they will eventually become unprofitable.
To address this looming profitability issue, Telecom Network Operators are trying to directly monetize content, apps and advertising. Doing so forces them to compete with “Over the Top” (OTT) providers like Google, Netflix, Amazon and Apple. Forcing a change here is difficult because the OTT providers are in a healthy position to defend their market position. They typically own the content, have a lower cost base through the use of standardized hardware, and without large legacy businesses they tend to be more nimble.
Assuming Telecom Network Operators continue to struggle to more effectively monetize content they will need to address the costs and inflexibility in their business model. Network Function Virtualization (NFV) has emerged to potentially help here.
Network Function Virtualization
NFV takes network functions such as VPNs, Intrusion detection, firewalls and load balancers and virtualizes them so that they can run on standard server hardware. In this way network functions can be built by independent software vendors and deployed on everything from bare metal hardware through to hosted cloud platforms. The diagram below contrasts the NFV approach against the current model dependent on dedicated hardware appliances.
There are several benefits to network function virtualization for the TNOs. These include:
- Increased service velocity – standardized interfaces allow new functions and application to be more quickly added to networks to create new service offerings.
- Optimized resource allocation – by building networks upon elastic environments there is no need to build for the worst case; the network can be configured on-demand.
- Reduced equipment costs – eliminating the need for specialized hardware should lead to greater competition amongst the providers of network equipment.
Of these benefits, the last two are relatively straight forward CAPEX and OPEX optimization techniques, whilst the first offers both the most potential and the most risk. If the TNOs can use NFV to more readily create new services of benefit to the OTT providers they should be able to move from competing with them to profiting from their services. The risk here is that migrating from a classical approach to virtual network approach will be complicated, both technically and commercially. So how might this play out?
How Might This Play Out?
On many fronts, NFV passes the disruptive innovation test – not as good as dedicated hardware but simpler, more convenient and much lower cost. As such, it has the potential to force industry leading network equipment vendors to chase the high-end with premium solutions, leaving the low end open for the disruption. This is the classic pattern for disruption.
However for NFV to gain widespread adoption, there are a number of challenges to overcome. These include:
- Portability - it will need to be portable amongst server hardware and cloud platforms.
- Modularity - allow different functions to integrate with one another including those from different vendors.
- Interoperability - co-exist and interface with existing specialist hardware that will remain in the network.
The key to overcoming these challenges lies in achieving a set of standardized interfaces between all parts of the virtualized and physical network. Currently NFV is being championed by a group of 28 tier 1 Telecom Network Operators in ETSI, the European Telecommunications Standards Institute.
If history is any indication, at some point the key network equipment manufacturers (NEMs), like Cisco, Alcatel-Lucent, Ericsson and Huawei, are going to attempt to assert their dominance and authority so that they can to continue to profit from specialized hardware. Varying tactics can be applied by the NEMs during the standardization process including asserting control through establishing first mover advantage, stalling and forming competing bodies. Whatever is employed, history shows each will work to steer the outcome to their advantage.
The hope for achieving standardized interfaces lies in the fact that NFV is defining requirements and shared architectural descriptions by which a virtualized TNO will operate, rather than following the norm and creating technical standards. Put another way, NFV is better viewed as the world’s largest RFP, as opposed to a standard. Consequently the TNOs and participating NEMs are collaborating in a highly interactive, and cooperative manner to develop the framework that will be used in the future by all suppliers to position their solutions for production operations.
In the best case, near term we are likely to see fragmented NFV implementations as NEMs and startups race for the first mover advantage. Several TNOs are already in the process of creating operational pilots ahead of the first round of stable documents which will be available in October 2013. Longer term, this is likely to be followed by a series of consolidations and delays as interoperability is reached. So whilst NFV looks promising on the surface, it is likely to be a long road to any commercial success.
NFV could prove to be a white knight for the TNOs, allowing them to co-exist, rather than compete with the OTTs through increased flexibility and lower cost of service creation. NFV does this by eliminating the need for the TNO to purchase expensive hardware to deploy a new service. Instead, they test new software and if the software works then great, if it doesn’t then they terminate the project. The cost goes from deploying hardware to deploying software in the NFV cloud. Clearly the later should be a fraction of the cost of the former.
Achieving this will require the formation of, and adherence to a standardized architecture. It is here that the NEMs are likely to slow the entire process down. In doing so NFV could turn from white knight to white elephant. Time will only tell how this plays out, however Citrix is in a unique position to influence the NFV requirements and shared architectural descriptions in what could be a revolutionary architectural shift to a multi-trillion dollar industry.