NFV in 2017: A Key Year for Commercial Deployment

[[181719]]

As mobile communications shift from the voice and data era to the traffic and large-connection era, operators' traditional network and equipment room construction and operation models can no longer adapt to future business development needs.

For China Mobile, under the development strategy of delaying the decline of voice services, increasing traffic growth and expanding digital services in new areas, IT has become the key to achieving network cost reduction, efficiency improvement, flexibility and openness, and value enhancement. Among them, NFV is one of the core technologies for IT development. In 2017, NFV will enter the key stage of commercial deployment, providing a new engine for network capability development and operation model innovation.

Comprehensive virtualization to build a new network

It has been more than four years since the concept of NFV was proposed. NFV originates from information technology and uses virtualization technology to decouple the software and hardware of traditional telecommunications equipment, realize telecommunications network functions based on general computing, storage, and network equipment, improve management and maintenance efficiency, and enhance system flexibility.

NFV is a major change in the field of mobile communications, and it can be applied in many fields of mobile communication networks. NFV can be applied in a wide range of fields, and the entire communication network equipment, including base station equipment, can achieve stronger capabilities through virtualization and software. Today's network evolution has gradually become virtualizing as much as possible and softwareizing as much as possible. There is a demand for virtualization from the access, control, core network, and business levels.

Based on NFV considerations, the three major operators have formulated new network strategies. China Mobile also formulated the NoveNet plan early on, namely new architecture, new operations and new services. Based on the NoveNet plan, China Mobile designed the future target network architecture into three layers. The bottom layer is similar to the recognized transmission equipment of the nearest kilometer. The upper two layers mainly exist in the form of DC, corresponding to the network functions above the current core layer, which are divided into two levels: core DC and edge DC. The core DC will try to concentrate the functions of the future control plane upward to realize a large centralized management mode. Functions related to user data and traffic will be sunk to the location closest to the user as much as possible, that is, the edge DC.

At the same time, in the new network architecture, China Mobile proposed the concept of Telecom Integrated Cloud, or TIC, as the core component of the future network architecture. In the future cloud network, by building new ones and upgrading traditional telecom rooms to TIC data centers, a nationwide network including core TIC and edge TIC will be formed to achieve the centralization of the control plane and the sinking of the media plane of the telecom network. It is hoped that in the future DC can achieve standardization and modularization, that is, end-to-end cloud plans that can be quickly replicated and deployed, and DCs can achieve unified WAN connection and link scheduling through SDN.

On top of these three layers, the orchestrator serves as the core brain, and through a unified collaborative orchestrator, the management, orchestration, and scheduling of networks, network elements, and services are realized, as well as the opening of capabilities to the outside world.

***The goal is to achieve three-layer decoupling

Virtualization deployment of core network functions still poses great challenges to the industry chain in terms of performance, stability, and management. From virtualization to cloudification, network reconstruction is the development goal of operators. In fact, the development of NFV has a bright future, but the road is tortuous.

One of the key issues is the three-layer decoupling. At present, there are many problems in achieving software and hardware decoupling. The three-layer decoupling involves vertical interoperability among multiple vendors, which makes system integration and maintenance difficult and takes a long time to deploy.

At present, China Mobile adopts a transitional solution of software and hardware decoupling, using X86 as the hardware, and the virtual layer and applications are packaged and provided by another manufacturer. However, the hardware management problem brought about by this has become an unavoidable key issue.

The first key issue is the server BOIS setting. Different virtual layer vendors have different requirements for hardware BOIS settings for computing nodes, and different vendors' servers have different default BOIS settings when they leave the factory. Another key issue is that interfaces are difficult to manage uniformly. For example, servers currently use IPMI and SNMP, disk arrays use SNMP and SMI-S, and switches use SNMP for management. Therefore, the standardization of hardware and VIM interfaces is a difficult problem that the industry urgently needs to break through.

However, this model is only a transitional solution. In the future, full decoupling of the three layers will inevitably be the target solution. The hardware, virtual layer and application are relatively independent to achieve cross-domain, global resource and network orchestration and scheduling. To achieve this goal, standards and open source will become the two major foundations. In the past, the industry mainly relied on standards, but in the future, both standards and open source will be very important. The decoupling of the three layers still requires long-term joint efforts of the industry.

Unified orchestration management and reconstruction of OSS

In terms of network management, Orchestrator will become the brain of future networks. It will not only enable end-to-end network resource management and business orchestration, but will also trigger changes in the network management OSS system to promote the formation of a new generation of network management and orchestration systems.

This view is raised because we have seen that as a core component of the NFV/SDN product system, the advantages of collaborative orchestration by Orchestrator are reflected in the management of the entire life cycle: for large-scale network creation, it can quickly generate and replicate networks based on network service templates; for on-demand network creation for customers, it provides on-demand service capabilities and realizes the creation and activation of customer self-service in a building block-like manner; for network upgrades/updates, unified remote software version updates for the entire network bring a new network upgrade experience; for network elasticity needs, based on pre-set strategies, it can realize flexible adjustment of network elastic expansion/contraction and disaster recovery backup under various business models such as the Spring Festival, activities, and emergencies; for network decommissioning, it can terminate network services on demand and recycle underlying resources.

In June 2016, China Mobile and well-known companies jointly established the OPEN-O project, the world's first unified SDN and NFV open source orchestrator. After the project was launched, the first version was released in November 2016 to help operators, equipment providers, developers, etc. align requirements more quickly and enhance multi-vendor interconnection.

At present, Orchestrator is still at the tool level and does not have intelligent features. In the future, Orchestrator will combine big data, artificial intelligence, etc. to realize dynamic and automatic orchestration and scheduling of the entire network. In the future, it is expected that NFV-O will be responsible for the orchestration of network functions, and will flexibly build network functions implemented by software to quickly provide services; SDN-O can realize centralized control of network connections, so that network connections can be automatically adjusted and optimized to adapt to rapid changes in business.

In addition, Orchestrator will have a huge impact on OSS. One is responsible for network resource scheduling and orchestration, lifecycle management, and the other focuses on network FCAPS management. The two must be closely integrated to collaboratively manage the network.

Therefore, the focus of future development will be on reconstructing OSS with Orchestrator. The short-term goal currently under consideration is to coordinate the FCAPS management and lifecycle management of network element/network service configuration, faults, and performance. In the long run, the goal is to maximize the business value of network virtualization. The functional focus is no longer limited to FCAPS management and NS lifecycle management. Through the collaboration of Orchestrator and OSS, business intelligent orchestration and intelligent operation and maintenance can be achieved.

In the future, there will be more end-to-end issues such as high reliability that still require long-term efforts, and the KPIs at the traditional business level will also be greatly affected by the introduction of virtualization. How operators can ensure the high reliability requirements of the five 9s of telecommunications networks and how to deal with the need to redefine related KPIs will also be further placed before the industry. (This article is based on Zhang Hao's speech at the 2016 China Communications Industry Annual Conference, with deletions and modifications, and the title is added by the editor.)