NFV needs to be commercially available on several fronts

NFV needs to be commercially available on several fronts

When talking about networks, there are two words that cannot be bypassed at all, namely SDN and NFV.

SDN emphasizes the separation of forwarding and control. By introducing a new control plane, it shields details of forwarding plane from different manufacturers and realizes intelligent and centralized network control functions; NFV emphasizes the decoupling between the hardware platform and business applications, and hopes to use cloud computing and virtualization to build a resource pool based on the standard hardware platform to break the chimney-style business deployment model in the past.

These two concepts are proposed with a certain degree of similarity.

The first is to reduce costs, hoping to replace the original dedicated hardware with white-board switches and x86 servers to reduce TCO, and the second is to get rid of supplier locks, both SDN and NFV emphasize openness and open source, giving users more choices, rather than using private protocols. The third is truly empowering the network and making the network smarter, enables quick activation and flexible deployment of services.

Particularly for carriers, SDN and NFV have more significance than that. They also hope to re-establish themselves to the core position of the industrial chain through the introduction of SDN/NFV, take the initiative in the game with OTT, or create new business models to meet requirements.

Since SDN and NFV have so many advantages, operators will certainly spare no effort to promote them. However, unfortunately, according to the current situation, the industrial progress is not very good. We can't help but ask, why did this happen?

1. conceptual challenges

This statement seems to be a bit difficult, but it is the first obstacle.

In my opinion, China Telecom is an industry and a kind of thinking. IT emphasizes the classification, control, and whole-process network. IT also comes from the virtualization and cloud computing concepts of the IT and Internet worlds, it emphasizes equality and openness.

These are two completely different ideas and ideas. In the early stages of telecom business development, its hardware capabilities and business models were able to keep up with requirements. However, the IP and Internet of telecom businesses have broken the original pace, the original method is already unsustainable. Businesses are all networked, and basic networks must also be networked and IT-based. De-Telecom is an inevitable choice.

However, over the years, industrial habits and ways of thinking have become solid. It takes time and courage to break this inertia. Sometimes, we have to pay too much for this "Evolution" from 3 to 5 9.

2. technical challenges

NFV will fundamentally change the deployment of telecom infrastructure and then the delivery of services by operators, but this also brings great challenges. I agree with HP's statement that these challenges can be categorized into three categories: infrastructure challenges, operation Challenges, and organizational challenges.

Infrastructure challenges: NFV is designed to replace dedicated hardware with standard hardware. It builds a flexible and scalable resource pool, and then fills in different business application software to implement different functions. The challenge here is whether the general x86 hardware server platform can meet the capability, reliability, and latency requirements of dedicated hardware, which are very important to telecom operators.

Here, I would like to praise Intel, the other end of the industry chain, because over the past 20 years, Intel has been constantly improving its chip processing capabilities and enriching its chip features. Taking DPDK as an example, this technology solves the forwarding capability issue well. Without DPDK, ordinary x86 chips cannot meet the needs of telecom operators in many scenarios.

The operator's business chain is clear under the original network architecture, but NFV introduces a new abstraction layer, which makes the existing BOSS pattern more complex, therefore, more orchestration work is required. How to maintain the customer and service view related to the underlying infrastructure, and ensure NFV agility and improve service speed through end-to-end automation. Another important operational challenge is how to ensure that NFV is deployed at the same time reduce operating costs-deploying NFV increases complexity, requires changing the lifecycle process, and requires association and automation at all levels.

The main challenge for organizations is how to break down the Department wall and change the existing process. Under the broad framework of NFV, how can R & D departments, informatization departments and network departments better cooperate and how should network planning and development departments and Procurement Construction departments carry out their work, whether it is necessary to introduce software development talents on a large scale or set up a CIO post similar to an enterprise requires consideration. But now it is clear that the way operators divide jobs based on transmission, wireless, network management, IP, and other specialties is changing.

3. Challenges of ecosystem and Model

To some extent, SDN and NFV break the original industrial framework, use open-source methods to lower the entry threshold of the industry, and innovate the original business model to stimulate the vitality of the industry. However, it takes time to cooperate with the industrial chain.

The telecom industry originally had a complete industrial chain, chip-system equipment-application software, and everyone had their own responsibilities. However, NFV gave many enterprises the opportunity to enter the telecom industry, with traditional IT hardware manufacturers, some software vendors and some start-up companies.

NFV makes the original Telecom Network feel more powerful and has lower barriers to participation. This requires a better industrial environment, rather than singing different voices. As the underlying NFV industry chain manufacturer, Intel is accelerating the maturity of OpenStack, DPDK, and OVS around open source and standard organizations. It is also actively promoting the growth of the Network Ecosystem, its Network Builders program has more than 180 members.

Business models are also a difficult issue for NFV commercialization. The original business model was very simple. Several equipment manufacturers were bidding, and the bottom bidder won the bid. If you do not open your business for one year, but NFV is different, the underlying cloud hardware price is almost transparent, where you can buy it, or even go to DIY. However, how do I charge for filling applications based on these hardware, such as DPI, firewall, and authentication services?

At present, there are several methods available: Sales authorization, which can be sold; service sales, authorization price is very low, after-sales service price is relatively high; patch sales, authorization price is very low, in subsequent version upgrades, fees are charged separately. In addition, the Service is sold as a whole and the total package is directly provided to the carrier.

In the words of Wang Feng, a Chinese telecom Technical Expert, software is a new thing for operators. First, the cost of the software cannot be calculated, and it cannot always be done according to the code line. Second, how many copies are sold after the software is developed, and the subsequent costs are shared, or even no money; third, the hardware can be used as an asset to enter the report. However, how to introduce the software that was originally used as a supporting service is a problem.