Produits, solutions et services pour les entreprises
By Wu Ping, Campus Network Product Management Director, Huawei
The core function of a campus network is to connect applications and terminals. In terms of applications, various application systems are being migrated from the local network to the cloud as enterprise ICT services become increasingly cloud-based and AI-powered. The cloudification of application systems, exponential growth of service traffic, and the widespread use of AI applications based on big data training are straining current campus networks. To keep up, future campus networks must provide high-quality connectivity that features larger capacity and lower latency. IoT technology is driving many terminals to go digital. As a result, the number of campus terminals soars along with data volume. In addition, wireless networks make terminal mobility a reality. Both of these drivers will revolutionize data generation, transmission, processing, and application, and greatly improve enterprise production efficiency. With insights into cloudification and AI adoption for applications, as well as IoT and wireless trends for terminals, we see four trends that will materialize in campus networks.
The wireless network is obviously crucial to improving enterprise productivity and efficiency. All-wireless networks are an inevitable trend. The last wave of wireless reconstruction mainly occurred in enterprise office and guest networks. However, the next wave of wireless reconstruction will happen in enterprise production networks, especially with the mature commercial use of the next-generation Wi-Fi technology — Wi-Fi 6. The production network has more stringent requirements for wireless transmission quality and reliability. Traditional Wi-Fi 5 is incapable of meeting these expectations. Wi-Fi 6, however, can make a difference, as it delivers four times the peak bandwidth and terminal access capacity and offers better anti-interference performance and lower network latency than Wi-Fi 5.
Wi-Fi 6 makes it possible for enterprise production networks to go wireless. This will open up new possibilities in scenarios such as intelligent manufacturing and AI-powered operations, which will greatly help enterprises improve production efficiency. We believe that Wi-Fi 6 APs will see large-scale commercial use and deployment in 2019. The consulting firm ABI predicts that the global shipment of Wi-Fi 6 will reach US$4 billion in 2021. Pleaser refer to ZK Research white paper-WiFi 6 Accelerates a Path to a Hyper-Connected World
Additionally, short-range low-power protocols (such as BLE, RF, and ZigBee) designed for the IoT will also be widely used in campus networks over the next 5 to 10 years. Driven by the all-wireless megatrend, future campus networks will integrate wired, wireless, and IoT services; fully connect everything; and remain always-on.
The convergence of wired, wireless, and IoT networks greatly reduces the CAPEX of enterprise network construction while raising higher requirements for network capabilities. Two types of services are emerging and will coexist on a single physical network, which will bring challenges to network vendors.
• The first type of service is the interaction between people. The sensing boundaries between the physical and digital worlds become increasingly blurred. Enterprises are pursuing digital communication as close to face-to-face communication as possible to improve production efficiency and enhance employee and customer experiences. Rich media applications represented by 8K videos as well as cloud Virtual Reality (VR) and Augmented Reality (AR) are requiring higher network bandwidth (200 Mbit/s per machine) and lower latency (end-to-end latency within 30 milliseconds and less than 10 milliseconds on the network side). We believe that these requirements will never decline, but instead will become ever more stringent.
• The second type of service is the interaction between objects and applications — that is, IoT services on the campus network. Campus IoT services vary with quality requirements in diverse application scenarios. Many IoT services may have less stringent requirements on bandwidth, but are highly sensitive to packet loss or delay. For example, we conducted joint innovation on an unmanned warehouse where Automated Guided Vehicles (AGVs) are used. During wireless roaming, if more than two packets are lost for an AGV, it will automatically stop, causing a series of chain reactions. As a result, all the other AGVs will run abnormally, negatively affecting services.
Undoubtedly, when two distinct services run on the same network, they will contend for resources and, as a result, conflicts may occur. Therefore, a large-capacity, high-reliability, and ultra-low-latency infrastructure must be used to prevent this. In addition, intelligent differentiated services must be offered to provide optimal network resources for each service. In this way, every service on the same network does not contradict others, and can deliver the best experience, even with limited resources. A future network system can continuously and digitally measure and evaluate user experiences and automatically adjust network resources through AI technology. As such, each service can obtain the optimal Service Level Agreement (SLA) assurance (such as packet loss rate, delay, and jitter). This autonomous and self-optimizing network will truly be a qualitative leap forward compared with the traditional network that statically deploys QoS policies.
Driven by heavy traffic pressure, the network architecture will constantly evolve toward simplicity and flexibility. This evolution is partly due to constant innovation, price-performance improvement, and the inevitable pursuit of high-quality services. The more complicated the system is, the more easily problems occur. In contrast, a more stable architecture means easier Operations and Maintenance (O&M) and management.
With the advent of Wi-Fi 6, wireless Access Points (APs) will deliver improved high-density wireless connection capabilities. Given this improvement, we start to pay more attention to improving the data rates of campus switches instead of increasing the number of interfaces. Currently, the wired networks of many enterprises are dominated by GE and 10 GE links. With the popularity of Wi-Fi 6, we recommend that these enterprises quickly upgrade to higher-speed links such as Multi-GE (2.5/5G/10G), 40 GE, and 100 GE.
With constant hardware upgrades in interface density and forwarding performance, the typical tree structure of campus networks will be further flattened into one similar to the spine-leaf structure of a data center — if the number of interfaces, over-subscription ratio, and physical cabling permit. The benefits of this flattened structure are greater flexibility, easier scalability, simpler management, and higher reliability. In addition, enterprise campus networks can easily evolve toward a nonblocking and lossless network without packet loss.
After the structure is simplified, the campus network must be more flexible so that it can provide powerful service capabilities. Flexibility is reflected in two aspects: easy capacity expansion and easy scalability. Specifically, the highly regulated network construction standards meet service development expectations over the next 3 to 5 years. Performance and applications are highly scalable thanks to the computing resource capabilities of campus network devices and modular, programmable, and open computing software architecture.
In recent years, growing security threats have emerged from the internal network (intranet). In particular, ransomware viruses are becoming more rampant, with many variants constantly surfacing. In addition, the convergence of the IoT and campus networks brings great challenges to campus network security. Due to their own limitations, IoT endpoints are easily spoofed, replaced, and poisoned with viruses.
We believe that traditional campus border security approaches are becoming inefficient. Advanced Persistent Threat (APT) attacks and intrusions into the intranet using encrypted traffic can quickly infiltrate enterprise intranets and threaten enterprise terminals and data. In response, enterprise administrators struggle to deal with these attacks or intrusions for several hours or even days.
Facing this, we need to introduce a new security solution to the campus network. On the one hand, we must build a zero-trust cyber security architecture. Intranet elements, such as switches, are the first line of security detection and defense, and they need to be deeply integrated with security capabilities. On the other hand, Network Elements (NEs), local security analyzers, network controllers, and intelligence centers on the cloud need to collaborate and build a ubiquitous, proactive security defense system.
