At 2023 MPLS SD & AI Net World Congress, Huawei demonstrated the L3.5+ data center autonomous driving network solution. This future-proof solution enables digital transformation of carriers and customers across industries, fuels the rapid growth of new services such as ChatGPT, 5G, and AI, minimizes the total cost of operation (TCO), and simplifies operations and maintenance (O&M) of data center networks (DCNs). All of this helps to accelerate data center (DC) evolution towards multi-cloud and multi-DC.
Currently, customers’ core requirement for DCNs is that they can guarantee rapid service innovation and rollout. This, however, is hindered by four challenges: multi-cloud, multi-vendor, multi-O&M view, and multi-work order.
To address these challenges, Huawei CloudFabric 3.0 offers the industry’s only L3.5+ data center autonomous driving network solution. The solution stands out with three key features:
- Rapid construction: The duration for setting up cross-cloud connectivity is slashed from months to minutes.
- Rapid deployment: Services are deployed and provisioned automatically, taking minutes instead of days.
- Rapid troubleshooting: The troubleshooting period is shortened from hours to minutes.
This feature-rich solution also implements full-lifecycle network automation spanning network planning and design, network deployment and provisioning, service rollout, network monitoring and troubleshooting, network change, and parameter adjustment for network optimization. The following figure shows the highlights of Huawei L3.5+ data center autonomous driving network solution.
iMaster NCE — the core component of the L3.5+ data center autonomous driving network solution — introduces the digital map technology to build a high-precision digital map for a multi-cloud, heterogeneous network. This enables network visibility from five dimensions — geographic location, service, application, topology, and device — and one-click service path navigation. All of this helps to build an all-digital network, laying a solid foundation for visualized business decision-making. Another trait of iMaster NCE is its undifferentiated second-level simulation capability, which can ensure 100% accuracy of cross-cloud heterogeneous network changes. The following figure shows the digital map graphical user interface (GUI) of iMaster NCE.
Another critical component of the solution is Huawei DC switch, which has outstanding performance in energy consumption thanks to its collaborative design of hardware and software. To elaborate, in terms of software, the switch is loaded with many innovative algorithms, such as intelligent lossless (iLossless) and network scale load balance (NSLB); as for hardware, the switch is equipped with vortex fans, phase change heat sinks, and a ventilation design with the industry’s largest ventilation hole. Such traits make it ideal for sustaining the green, low-carbon development of DCs. In the AI training scenario, the NSLB algorithm enables global load balancing, improving effective network throughput to 90% and AI training efficiency by 20%. The resulting benefits include 13% lower energy consumption. The following figure compares the NSLB algorithm with the traditional load balancing approach.
Thanks to its differentiated advantages in technological innovation, Huawei L3.5+ data center autonomous driving network solution is favored by carriers and customers across industries, and has seen significant uptake worldwide. Looking to the future, Huawei will continue to make efforts in exploring with industry partners to deepen the construction of the autonomous driving network evaluation system, guide network upgrade towards automation, intelligence and intergenerational network evolution, and accelerate the digital transformation pace of carriers and customers across industries, thereby helping our customers to achieve business success.