It has been a popular buzzword to add abbreviations SDN/NFV (software define network / network function virtualization) in any telecom presentations during the last years. For me the NFV hit hard around 2011 when showcasing opportunities in Cebit 2011. It was mind blowing to explain how you can increase network capacity in a couple of minutes instead of months and just by clicking your computer. This is definitely one of the benefits of NFV, but it is the most significant.
Telecom has been and still is a deep pocket game. Telecom products have traditionally been described with an adjective, telecom grade. This has meant 99.999% reliability. Telecom grade is the opposite term to IT product, the world where reliability has been considered to be less important. This world was left behind some 20 years ago, but the change has been slow. Traditionally reliability of telecom products has meant custom designed components and redundant HW designs. This has meant low volumes and high prices. Nowadays the claim is more on the performance side, we need something special to meet delay requirements.
For the telecom ecosystem designing custom made HW means very high costs (design and product) meaning that there are not that many companies able to do this. Very good example is comparison between WLAN AP and LTE pico eNodeB. They are basically the same function devices, but eNodeB cost is some 5x more. WLAN AP is manufactured by hundreds of companies, LTE picos about 5. WLAN is not considered telecom grade, but still about 75% of internet traffic uses WLAN as an access media. WLAN is used in hospitals and used for banking, just giving some examples. Now with NFV we can get rid of telecom grade HW, not for RF yet but for the rest of the network. When network functions are based on general purpose HW and open source platforms any size of company can build lets say switches, database, policy functions etc. In telecom everything is fairly standardized and new standards are published on a regular basis so it is straightforward for smaller and agile companies to enter the market place. This will mean more competition and significantly lower prices. When there are more vendors the role of integrator will increase. Here is an interesting analogy with retail, department stores are in trouble because smaller vendors can sell directly to consumers.
SDN then has given us flexible networks. It is not any more needed to send somebody to the field connecting cables (well some labor is needed to set-up networks) but you can actually do configuration changes from your desk. There is a lot of functionality that can be taken even further, let the customers do their configurations by themselves. Networks are capable of being flexible, but there is a problem with user interfaces and orchestration in general. Changes are coming gradually but steadily. As a comparison you can think how stock trading is done today and compare that to the style some 20 years ago. There is also a lot of room for small and focused companies to build products on top of SDN capabilities.
To summarize, NFV is about CAPEX changes. To fully utilize its promise new competition needs to be introduced to the market. If NFV is left only for traditional telecom OEMs, cost saving potential is not fully utilized. SDN is about OPEX savings. We need to set bar higher than today and target customers to do their own configurations. This is again an area where small and agile companies can deliver new kinds of services. You might wonder if there is still a need for big telco OEM giants, yes there is a need for malls, their business model just needs to be changed a bit from their own designed and manufactured house to integrator role.