Traditional metropolitan area network architecture encounters bottlenecks and introducing NFV becomes an effective solution

With the rapid development of "Internet +", Internet facilities and applications are becoming increasingly rich, network traffic is growing explosively, the network is becoming more and more complex, and the operation and maintenance costs are rising accordingly; correspondingly, users' types and requirements for network service quality are constantly increasing, and the differentiated demands for not only access bandwidth but also service types and security levels are also growing sharply. The traditional way for telecom operators to cope with the growth in demand is to deploy different dedicated equipment for various services and vertically expand the original equipment for the growing services. This deployment idea is mainly suitable for situations where there are not many types of services and the new deployment demand is mainly for the expansion of similar services.

However, with the emergence of new Internet services, the original deployment ideas are obviously unable to meet the needs of various services for bandwidth resources and launch speed. As the carrier of broadband Internet services, if the metropolitan area network simply expands the network bandwidth instead of providing flexible usage methods, the edge nodes of the metropolitan area network will be piped.

The industry generally believes that by introducing NFV (Network Function Virtualization) technology, all or part of the functions of network equipment can be unified on high-performance, large-capacity servers that meet industrial standards, thereby reducing expansion costs and operation and maintenance costs while improving the ability to deploy new services and ensure segmented services.

Based on the current status and future needs of metropolitan area networks under the new situation, this article proposes a metropolitan area network introduction solution based on NFV technology, which will help solve problems such as low flexibility of edge nodes in operators' traditional networks, complex operation and maintenance, and difficulty in introducing new technologies.

There are five major problems with existing metropolitan area networks

A typical metropolitan area bearer network is divided into a user layer, an access layer, an aggregation layer, a service control layer, and a core layer. Its basic networking mode is shown in Figure 1.

Figure 1 Typical architecture of a metropolitan area network

The user layer is mainly composed of users' broadband access devices, such as STB (Set Top Box); the access layer provides access functions for user devices; the business information flow is aggregated through the aggregation layer; and then most of the core functions of the metropolitan area network edge nodes such as user logic information processing, configuration distribution, and user authentication are undertaken by the business access control layer. Typical devices are BRAS (Broadband Remote Access Server), BNG (Broadband Network Gateway), SR (Service Router), etc.; the core layer is mainly responsible for connecting with the metropolitan area backbone network and providing high-speed switching capabilities.

As the penetration rate of services such as video access and high-speed browsing on the user side increases, various new services are emerging. The common problems in metropolitan area networks at this stage include the following five aspects.

• Forwarding control capability mismatch: When the forwarding capability utilization rate of the device has reached 80%, its control capability utilization rate is only 20%;
• Insufficient network capacity: The existing network capacity is insufficient to support the rapid development of business;
• Inability to achieve flexible deployment: Existing dedicated hardware devices such as BRAS/SR/BNG have problems such as inflexible deployment and long construction period;
• Unable to flexibly respond to large traffic bursts: Lack of traffic scheduling capabilities for business resources;
• Complex operation and maintenance: With the growth of broadband services, many manufacturers and equipment types are involved. The maintenance interfaces of equipment from different manufacturers are not interconnected, and network equipment cannot be centrally controlled.

Different virtualization technologies have their own advantages and disadvantages

Virtualization means that computing units are run on a virtual basis (rather than a physical basis). It is a solution to simplify management and optimize resources. By introducing a virtualization layer, real physical resources can be managed downwards and virtual system resources can be provided upwards. Virtualization can improve the utilization of physical resources, strengthen the management of resources by managers, and improve the security and reliability of the system.

System virtualization is a type of virtualization technology, which refers to the virtualization of a physical computer system into one or more virtual computer systems. This technology abstracts a single physical platform into several virtual platforms through the virtualization layer. Each virtual platform has its own virtual hardware, providing an independent execution environment for the operating system. According to different virtualization architectures, virtualization can be divided into three categories: Type I virtualization, Type II virtualization, and operating system-level virtualization.

Type I virtualization is bare metal virtualization, where the virtualization layer runs directly on the hardware; Type II virtualization is host architecture virtualization, where the virtualization layer runs on the host operating system; OS Containers is operating system-level virtualization, where the virtualization layer is integrated into the host kernel. In actual application deployment, Type I virtualization or OS Containers virtualization architecture is generally used.

Depending on the degree of virtualization, it can be divided into full virtualization (corresponding to Type I virtualization or Type II virtualization), hardware-assisted virtualization (corresponding to Type I virtualization) or paravirtualization (corresponding to OS Containers virtualization). The comparison of the above virtualization architectures is shown in the table.

Table Virtualization Technology Comparison

NFV is based on the application of virtualization technology on network equipment. In the current network deployment, hardware-assisted virtualization or paravirtualization technology is generally used. Its core purpose is to unify the hardware of communication equipment into server equipment based on X86 architecture to reduce costs. Different functions (such as BRAS, etc.) can be realized by injecting different software on it.

After the service control layer equipment is NFVized, its resource utilization and flexible management can be improved while reducing equipment costs. Based on the software infusion mode, new services can be quickly deployed, accelerating the evolution of service control layer equipment towards pooling and cloudification. Therefore, NFV is a better evolution direction for metropolitan area network service control layer equipment.

Deploy intelligent metropolitan area network edge nodes based on NFV

Figure 2: Metropolitan area network edge node architecture based on vBRAS resource pool

The typical networking architecture of vBRAS is shown in Figure 2. The vBRAS device is deployed in the resource pool mode. It is deployed at the metropolitan area network service control access layer, connected to the core router above and the aggregation layer equipment below. The resource pool is mainly divided into the vBRAS forwarding pool and the vBRAS control pool, which are mainly responsible for the forwarding and control of the vBRAS. The forwarding pool is connected to the core layer and access layer equipment.

Considering that at this stage, the business control layer equipment of the metropolitan area network is mainly based on traditional BRAS/BNG, operators can give priority to introducing vBRAS equipment in local hot spots, and use vBRAS to quickly and flexibly go online to share traffic. That is, the user's large-granular traffic can interact with the core router through the traditional BRAS equipment on the left side of Figure 2, and the small-granular flexible traffic (such as traffic value-added services) is forwarded through the vBRAS equipment. At the same time, when the traditional BRAS equipment fails, the vBRAS resource pool can act as a temporary disaster recovery device to prevent user business interruption.

From a deployment perspective, the metropolitan area network edge considers introducing VxLAN-SW bypass or replacement in local areas:

1. The original SW forwards the services that need to be carried by vBRAS to the VxLAN switch according to VLAN;
2. The VxLAN switch maps the original VLAN service to VxLAN UDP packets;
3. UDP packets traverse the metropolitan area network to the resource pool of the vBRAS, achieving large-scale Layer 2 intercommunication.

vBRAS initially adopted a serial connection method to realize the cloudification of BRAS functions, and gradually introduced SDN in the later stage. The controller was centrally deployed in the network cloud, and the traffic was forwarded locally. In the future, operators can apply vBRAS equipment based on NFV technology to the following scenarios.

• Flexible traffic diversion and disaster recovery: When the traditional BRAS capacity is insufficient or fails, the service can be quickly diverted to the vBRAS resource pool through another path through VxLAN;
• Flexible development of personalized needs: Users with personalized needs can obtain services through the vBRAS resource pool, and the network supports flexible adjustment capabilities of the large second layer;
• vBRAS gradually integrates NAT, CDN, DPI and other functions to become an edge intelligent node.

summary

With the gradual opening of the telecommunications infrastructure construction and operation market, the competition in broadband network operation has become more intense. It is urgent to introduce new technical means to ensure the service quality and sustainable development of broadband networks, so as to adapt to the new development situation brought about by market competition. This paper studies virtualization and NFV technology, and proposes a deployment plan for vBRAS based on the current situation of metropolitan area networks, which also brings a new idea for the development of network technology.

Author: Ao Guili, China Mobile Communications Group Guangdong Co., Ltd.

Zeng Jielin, Guangdong Telecommunications Planning and Design Institute Co., Ltd.