Forecasting the NFV Opportunity

Network functions virtualization (NFV) has the potential to transform the way telecom networks are built and operated. Key questions remain about the leading use cases, market drivers, standards, and timing of NFV deployments.

NFV aims to transform the way that network operators build networks by using standard IT virtualization technology to consolidate many network equipment types onto industry standard servers (COTS). Leading communications service providers initiated a NFV standardization effort in the European Telecommunications Standards Institute (ETSI) in the fall of 2012.

As of mid-2013, standards efforts and proofs-of-concept are ongoing, but still in the early phases. Understanding the opportunity for a market which is just now starting to develop requires the analyst to make a number of assumptions about the future.

According to the best-case market scenario from Doyle Research, the NFV market will start to gain momentum with early use cases during 2014. The market is forecast to reach approximately $5 billion by 2018, including software, servers, and storage.

Here's some insight into how that number is determined. In order to properly discuss the market size of the NFV market, we need to think about the following questions:

  • Market definition: What is included in the NFV market? Software, hardware, virtual functions, and professional/support services can all be considered here.
  • Use cases: What are the leading NFV use cases (machine-to-machine, CPE, hybrid cloud, etc.) and how will they impact various parts of the existing network infrastructure? How will these use cases drive NFV deployment time frames?
  • Network equipment in scope: What types of network equipment will be in scope for NFV technology in what time frame? Leading types of equipment under consideration for NFV include the mobile core, IP Multimedia Subsystems (IMS), policy/service assurance, deep packet inspection, WAN optimization, security elements, some routing elements, and many others.
  • Key benefits: The key benefits that communications service providers will derive from NFV implementation include faster time-to-market, enablement of new services, ability to rapidly scale resources up and down, and lower costs (both CAPEX and OPEX). Which of these key benefits (or combinations thereof) will motivate CSPs to make the significant changes required to migrate to NFV?
  • Inhibitors: CSPs have large installations of highly reliable, revenue-producing networks. How will NFV impact the existing network? What are the risk factors? Who will provide the expertise to enable NFV migration?
  • Technological maturity: Some parts of the telecom network (mainly the application and some control plane elements) already run on standard servers (COTS). Other parts of the network (e.g., optical transport, routers, mobile edge, and switches) drive their scale and performance from a combination of proprietary ASICs or network processing units with integrated software -- making them difficult to virtualize. How well will NFV technology scale and how reliable will it be in the various parts (functions) of the network?
  • Standards/interoperability: The key advantage of NFV is the ability of CSPs to mix and match various software elements on standard COTS hardware. How well will various NFV solutions interoperate? How easily can they be tied to existing OSS/BSS (back office) systems?
  • Speed: How quickly can CSPs change the relationship between their network and IT operations? In order to broadly implement NFV, large traditional CSPs need to change their current organizational structure. They will have to blend expertise from their IT, cloud, and network organizations. Very few CSPs (and none of the large ones) have this capability today.

Having analyzed the wide range of questions outlined above, the market analyst needs to clearly define forecast assumptions for his/her best-case scenario regarding NFV adoption. All NFV forecasts recognize the nascent stage of the market, but all will differ on definitions, timing, technology, and change assumptions.

— Lee Doyle, Principal Analyst, Doyle Research

Luiz Lourenco 9/2/2013 | 10:23:58 AM
Re: Mobile edge? Actually operators all over the world are implementing IP RAN (end-to-end IP) solutions aiming exactly on the possibility to fast deploy new services and share the unused capacity with other networks. A new IP/MPLS RAN can be used to transport 2G, 3G and 4G/LTE traffic and also carry the packets from UBB and other FTTx access services, all simultaneously. An IP/MPLS network can have many of its functions virtualized and centralized based on the SDN and NFV principles, so we're likey to see a good part of the mobile edge and RAN running on virtual servers in the near future, I mean 3 to 5 years ahead.
Dredgie 8/28/2013 | 3:59:44 PM
Re: Mobile edge? Virtualization of the mobile base station has already been identified as an early use case for NFV – specifically by members of the carrier community. In all fairness, the NFV work, here, is just an extension of the cloud/centralized RAN (C-RAN) research project initiated by the China Mobile Research Institute. This initiative serves to reduce the general underutilization of RAN that must be architected for peak hours with capacity that cannot be shared.

As you mention, there are baseband processing functions that may not be efficiently realized in software. Remember, however, that in many cases, NFV is finally forcing the issue of control and media plane separation. In this instance, there could be a base band unit (BBU) resource pool serving virtualized eNodeB's, Home eNodeB's, Femto and Pico cells. Obviously, there's work to do, here, in terms of northbound interfaces to the virtualization orchestrator, etc...
sam masud 8/28/2013 | 1:21:16 PM
Re: Mobile edge? Agree. I think putting a dollar estimate on the NFV market would be pretty tricky because not all network functions are going to get virtualized at the same time.
bcohen777 8/28/2013 | 12:42:16 PM
Please Show Us the Data!! I second Carol. It would be wonderful to have a simple, self-explanatory chart documenting the main areas included in the estimate -- servers, software, etc. and some explanation of any special concerns with coming up with the answer.
LeeDoyle 8/28/2013 | 11:12:20 AM
Re: Mobile edge? All elements of the telecom network including wireline, wireless, and cable are POTENTIALLY in scope.  I do not include enterprise (business) networks as party of this NFV analysis.
Carol Wilson 8/28/2013 | 11:01:17 AM
Re: Mobile edge? So Lee, what do you include in the NFV market? As you say above, it's not all that well defined and could include almost any part of the network. 
LeeDoyle 8/28/2013 | 10:54:14 AM
Re: Mobile edge? Cloud RAN is an interesting area for NFV.  I think operators will experiement with moving some of the RAN intelligence into the cloud, but the RAN itself is likely to remain specialized integrated hardware.
DanJones 8/28/2013 | 9:43:43 AM
Mobile edge? So how much of the mobile edge of the network can actually be virtualized, given the custom chips and other non-standard elements there?
Sign In