Development Trends for Router Technologies

Over the past 30 years, IP networks with routers and switches have developed according to Moore's Law and have been critical to the emergence of a modern information society. Being a core for WAN and LAN outlets, routers have played a key role and experienced sustained development and numerous upgrades. Router interfaces have developed from a wide variety of narrowband interfaces, such as X.25, E1, ATM, POS, and SDH, to a majority of broadband Ethernet interfaces and a few high-performance and high-speed WDM interfaces. Protocols for the interfaces have also developed into distributed IP/MPLS architecture. Routers and switches now have higher bandwidth and shorter delay, and they are also smaller and cheaper. They have therefore become the basis of almost all networks.
As change and motion is eternal, development is critically important. Being affected by 100M FTTH broadband, online 4K videos, 4G/4.5G/5G, IoT, the idea of user experience first, as well as the concepts of software-defined networking (SDN) and network function virtualization (NFV), routers are also facing a continuous evolution and change.
Router-based IPWANs are used widely. Because it is expensive to build a new network, most operators focus on network expansion and technical compatibility. Therefore, a basic development trend or competition focus for routers is still an evolution at an appropriate speed. Their capacity and capabilities are continually improved by inheriting existing functions. Currently, a mainstream commercial core router has 16–20 slots in each system and can form a 6.4–8T subrack system with 400 GB for each board. In future development, 1T, 2T and 4T boards will be launched to increase capacity without increasing power consumption or price. This sounds simple, but the industrial supply chain has been nearly destroyed and needs to be vertically integrated. In this context, launching products with appropriate capacity and costs at the right time is a big test to a company's overall strength that involves new technology reservation and IC design capability especially NP design capability. This is almost equated with technical competitiveness of high-end routers.
Another trend is smaller size. Traditional mid-range and high-end routers seem to be bulky iron monsters. As property costs increase year after year, small high-bandwidth routers have become very popular. Some operators even expect outdoor pole-mounted base station routers. It is therefore important to research compact outdoor routers. Such routers will definitely be favored by customers and may bring differentiated competitive advantages in the market dominated by traditional router vendors. Close attention must be paid to smaller size as an important direction of router development.
A third trend is customer experience. MI home gateway routers bring good user experience and their design is quite recommended. The hardware of MI routers is made up of common elements, but its software creates an app-based easy-to-use application chain that can create value and promote sales of more MI smart home appliances. ZTE focuses on mid-range and high-end routers and can also borrow this concept to migrate internet elements to its routers to create a convenient eco-environment and enhance customer loyalty.
The three trends are within the scope of router evolution. With the development of SDN and NFV, router-centric solutions are also undergoing a revolution.
The introduction of SDN is a basic point of the revolution. Deploying IP networks worldwide relies on a distributed protocol system where IP networks can still operate smoothly if one or more routers fail. However, the distributed protocol system makes it difficult to implement global special strategies and lowers the intelligence of IP networks. Moreover, the break in network control between the application layer software and router software also makes it difficult to automatically implement application strategies.
SDN solves these problems. Strategies for introducing SDN into WANs, where routers are widely used, are comparatively conservative. Few changes are made on forwarding planes of the routers, and control planes should be compatible. A hybrid system uses controllers to gradually centralize operations of the control planes previously distributed on the routers. Through hierarchical controllers and appropriate northbound interfaces, complete software automation is possible. The controllers use original interfaces, such as NETCONF and CLI, or extended interfaces, such as BGP-LS and PCEP, to connect devices. The original PW, L3VPN, and MPLS TE tunnels also remain on routers, but the operation mode is changed. SDN has significantly addressed the low intelligence issue with IP networks. In the longer term, although the current evolution is relatively conservative, the pace of evolution continues and generalizing the forwarding planes will become a new research direction after the control planes become SDN-based.
SDN is just a starting point of the revolution, and NFV brings about more changes. NFV aims to generalize functions. Previously, network service functions such as NAT, firewall, DPI, and load balancing were activated by special devices or boards. These special devices or boards are networked with the forwarding routers to provide intelligence or application layer (layer-7) functions. NFV has dramatically changed networking. Functions at all service layers including IP/MPLS forwarding can be virtualized as software running on the servers, which can completely eliminate routers. Yet these functions can also operate on new routers through accelerated optimization in a virtualized or non-virtualized manner, which provides new routers with an unprecedented and full intelligence and makes them adapt to different application scenarios. This game depends on the depth of thought and practice and will eventually achieve a balance through the deployment on demand. The functions on servers in a data center will be fully virtualized, and enterprise outlets will operate on new integrated routers. These servers and routers can share software between themselves, but they must be deployed for better customer experience, helping customers solve their own problems. Any idea that truly creates value for customers will be respected.
NFV-based routers will also be increasingly important in future network architecture, where data centers are more important. These routers will easily help network outlets of virtual data centers to provide functions for access to public networks or internet.
The development and application of NFV will enable universal services and integrated forwarding hardware on routers. This makes it easier to implement a SFC service chain and also gradually changes SDN forwarding planes and may even lead to the second revolution of IPWAN SDN. Though it is not quite clear when the revolution will occur, what is important is that the thought and practice on SDN and NFV that focuses on applications to truly create value for customers needs to be actively carried out.
Routers will be important in future networks. Focusing on customer value and experience, ZTE will continually evolve and optimize its network solutions, and be actively involved in SDN and NFV innovation to create new value and obtain better ROI.