Leading the Intelligent IP Era

Huawei’s vision and mission are to bring digital to every person, home and organization for a fully connected, intelligent world. Inspired by this goal, we believe that data communications networks are the foundation on which everything can be connected.

We can think of data communications networks as a series of waterbodies. As such, IP transport for 5G and fiber broadband networks can be analogized as brooks and streams, with numerous small tributaries finally converging into a parent river. Furthering this analogy, enterprise campus networks are comparable to ponds, with diversified services and varied connections, while IP Metropolitan Area Networks (MANs) and backbone networks are like huge, gushing rivers. Data Center Networks (DCNs) can be imagined as lakes, with some super DCNs like oceans, to underpin the connectivity, computing, and storage of vast volumes of data. Just as rivers, lakes, and oceans are endless, so will network connectivity be boundless and the intelligent world be limitless.

In the earliest days after inception, data communications networks mainly served applications such as email and bulletin boards. At that time, service experience of Internet users was deemed to have been fulfilled as long as there was network connectivity.

As enterprises deepen their digital transformations, their networks undergo new changes, such as a growing number of applications, ever-increasing network scale and complexity, higher network bandwidth demand, and increased network traffic. These changes bring unprecedented challenges to network planning, deployment, maintenance, and optimization.

So, how should we address these new challenges? A viable solution is to introduce digital transformation into the networks and leverage the technology bonus to cope with the challenges. To this end, Huawei takes the lead in introducing big data, AI, and next-generation protocols into data communications networks, bringing data analytics and closed-loop feedback for optimization. This is a giant leap forward from traditional IP networks to intelligent IP networks. In 2019, Huawei officially released its intelligent IP network architecture and next-generation data communications products powered by AI Turbo technologies, leading data communications networks into the intelligent IP era.

Three Characteristics of Intelligent IP Networks

Super capacity

The network bandwidth demand keeps increasing with surging traffic. Accordingly, campus networks are upgrading to Wi-Fi 6 and 100GE switches, while data center networks and IP backbone networks are moving towards 400GE interfaces. Huawei’s data communications network products and solutions constantly set new industry benchmarks in bandwidth and performance, capitalize on engineering breakthroughs in material and system engineering innovations as a result of long-term R&D investment.

For example, the flagship data center switch CloudEngine 16800 comes with the industry’s highest density 48-port 400GE line card, while the AirEngine Wi-Fi 6 AP stands out with an all-time high bandwidth of 10.75 Gbps. Huawei’s NetEngine 8000 high-end routers provide the industry’s highest-density line card (36 x 400GE).

To deliver greater flexibility and efficiency in bandwidth usage, Huawei is the first vendor in the industry to propose and develop the FlexE-based network slicing solution. Similar to dedicated bus lanes on a road, this solution implements strict bandwidth isolation between different types of service traffic on the ultra-broadband network. In this way, this solution provides 100 percent committed bandwidth for critical services of vertical industries (such as 5G telemedicine), production networks in enterprises, and carrier IP private lines, and allows flexible adjustment of slice bandwidth to cater for service changes.

Intelligent experience

Services and applications have varied requirements for network connections. Typically, video conferences require real-time connection services, production control systems call for highly reliable connection services, and data backup needs cost-effective connection services. Users often complain about poor service experience, while overlooking a very important fact: To obtain good service experience, the network must first understand service intents. Only then can the network provide the desired experiences of connection services in line with service intents.

But how does the network obtain service intents? With traditional networks, this is not possible. With AI inference capabilities and service models from previous AI training, the intelligent IP networks will analyze packet header sequences of various service flows in real time, identify service types and finally derive service intents from service types. For example, in an AI computing data center, AI training traffic flows can be recognized by analytics because packet header sequences of AI training traffic flows have distinct characteristics. The network then knows that zero packet loss is expected by those traffic flows.

How can we better serve the known service intentions? On traditional networks, service provisioning is static and therefore unable to configure network resources in real time based on service intents, let alone quickly adjust network resources upon network status changes. In this context, insisting on traditional approaches cannot realize service intents or ensure service connection experience. This is where intelligent IP networks come in. By using telemetry and AI technologies, intelligent IP networks make it easy to analyze and predict the network status in real time and adjust network resource configurations accordingly to continuously realize service intents and ensure connection experience.

For example, to ensure optimal connection experience in site-to-cloud connectivity scenarios, quick network adjustment upon cloud changes is required so that WAN resources can be configured and adjusted in real time based on the network connection requirements of cloud services. In the past, routes were statically selected based on network topologies for WAN connections, making it impossible to perform intelligent traffic steering according to real-time congestion status. Huawei is the first to provide an SRv6-based solution that allows optimal paths to be intelligently selected and adjusted in real time based on service intents, network topologies, and network congestion status to continuously provide optimal connection experience.

When it comes to DCNs, where lossless network connections should be provided for AI training traffic flows, despite frequent traffic fluctuation, Huawei’s data center switches can dynamically adjust the buffer thresholds of the forwarding line cards in real time, based on service intents and AI-powered prediction, continuously providing lossless connection services.

Autonomous driving

The ever-growing network scale and complexity bring great challenges to today’s complaint-driven “post mortem” style network O&M, where the network O&M department is often the last department to know that a network fault occurred. So, how can we quickly rectify faults or perform predictive O&M? An ideal solution is to introduce the autonomous driving system, a key differentiator in intelligent IP networks. This new system can detect the network status in real time, quickly identify network faults or potential issues, accurately identify root causes through fault model matching, and automatically rectify faults or issues. In this way, the impact on user experience is minimized or even eliminated, network faults are greatly reduced, and network O&M efficiency is increased dramatically.

Working together with customers, Huawei has actively engaged in and explored autonomous driving networks. In particular, Huawei has proposed a five-level autonomous driving network evolution standard that helps ecosystem partners jointly explore the evolution path toward autonomous driving networks.

For example, in the DCN domain, Huawei and leading customers have made great strides in joint innovations—detecting 75 types of typical faults within one minute, locating them within three minutes, and rectifying them within five minutes. This groundbreaking milestone marks Huawei as the first to achieve Level-3 autonomous driving networks in the DCN domain.

Joint Innovation with Customer Pioneers to Lead the Intelligent IP Era

Together with industry pioneers, Huawei has carried out a series of joint innovations on intelligent IP networks. These concerted efforts have paid off, achieving numerous productive innovation outcomes.

Huawei’s CloudCampus solution, ideal for campus networks of all sizes, stands out with Wi-Fi 6 products powered by Huawei 5G and the industry’s highest-density 100GE campus switches, delivering gigabit ultrafast connection experience. This feature-rich solution also uses AI technology to realize intelligent O&M, troubleshooting 85 percent of typical faults in minutes and greatly improving the reliability of campus networks. Furthering this, intelligent prediction of terminal trajectories enables lossless Wi-Fi roaming and improves service experience. All these differentiators help lead campus networks into the intelligent gigabit-wireless era.

Huawei’s CloudFabric solution transforms DCNs to take on three new features: ultra-broadband, lossless, and full intelligence. This future-proof solution fully supports high-density 400GE line cards and uses iLossless — the innovative intelligent lossless algorithm — to fully unleash AI computing power. So far, Huawei is the only vendor in the industry to implement Level-3 autonomous driving networks, leading DCNs into the intelligence era.

Huawei’s CloudWAN solution implements strict isolation of office and production services through network slicing and enables real-time, dynamic adjustment of service paths via SRv6. This ensures converged transport of all services, including office and production services, and delivers guaranteed experience, leading WANs into the all-service intelligence era.

In 2019, Huawei took the lead in unveiling the one-of-a-kind HiSecEngine series AI firewalls. Equipped with a threat detection engine based on AI chips, these firewalls can detect unknown threats at an accuracy of up to 99 percent while delivering five times the industry average performance. These innovative AI firewalls lead network security into the intelligent defense era.

Because of the public health crisis in 2020, we are becoming accustomed to new application scenarios such as video conferencing, collaborative office, wireless projection, and robot services. New infrastructure deployments, covering 5G, artificial intelligence, cloud computing, and industrial Internet, are also getting into full swing. These new changes are posing higher requirements on data communications networks. Against this backdrop, intelligent IP networks are becoming the natural choice in the new era and will also be an inevitable trend for data communications networks for the next 10 years. We firmly believe that 2020 will be the first year for large-scale commercial use of intelligent IP networks.

Huawei stays customer-centric and keeps innovating to create greater value for customers. We join forces with industry players to accelerate the in-depth integration of AI into networks and jointly build intelligent IP networks, leading the intelligent IP era.