China Mobile and ZTE Jointly Launch Scale-Across Solution

China Mobile and ZTE Jointly Launch Scale-Across (GSE-DCI) Solution, Solving Pain Points of Ultra-Large-Scale Computing Power Coordination

China Mobile and ZTE have jointly launched the Scale-Across (GSE-DCI) solution, solving industry pain points such as low transmission efficiency, low resource utilization, and low system stability in long-haul interconnects by integrating key technologies including SuperPipe and Fast CNP
With in-house developed 800G chassis network equipment as the hardware foundation, the end-to-end Scale-Across solution was first verified on a 800G platform, achieving high-bandwidth, low-latency, and high-reliability cross-domain computing power coordination, and setting a benchmark for building intelligent computing networks

ZTE and China Mobile announced that they have jointly launched the Scale-Across solution.

Focusing on Data Center Interconnect scenarios, the solution solves industry pain points such as low transmission efficiency, low resource utilization, and low system stability in long-haul interconnects by integrating key technologies including SuperPipe (wide-area packet spraying) and Fast CNP (fast congestion notification). This marks a significant breakthrough in the field of collaborative innovation for intelligent computing networks, and provides a vital technical pathway and practical implementation blueprint for addressing the global challenge of cross-domain computing power coordination.

As AI technologies rapidly evolve, the parameter scale and application scenarios of AI foundation models continue to expand, driving an exponential surge in computing power demands. A single intelligent computing center, constrained by local computational resources, energy supply and network bandwidth, cannot efficiently support the distributed training and real-time inference tasks of ultra-large-scale models. Consequently, the need for cross-regional data coordination and multi-center computing power collaboration is becoming increasingly urgent. In this case, the evolution path of computing infrastructure is extending from traditional Scale-Up and Scale-Out to Scale-Across, enabling efficient cross-regional computing power coordination and global intelligent scheduling.

The equipment integrates cutting-edge technologies such as the orthogonal CLOS architecture, cell switching, and virtual output queues to provide a high-bandwidth, low-latency, and high-reliability hardware foundation for cross-domain intelligent computing.

With a number of technological advantages, the solution fully match the demands of ultra-large-scale intelligent computing centers and DCI networks.

Highly scalabe architecture for smooth evolution: Featuring an orthogonal CLOS architecture, each slot supports up to 36x800G ports, reserving sufficient capacity for future expansion and upgrades of intelligent computing centers and DCI networks.
Long-haul lossless transmission for guaranteed performance: Equipped with large buffers and a fast congestion notification mechanism, it significantly reduces latency and packet loss caused by long-distance transmission, ensuring high throughput and stability for cross-center intelligent computing services.
Intelligent traffic scheduling for optimized resource utilization: Leveraging the SuperPipe technology, it supports packet-level dynamic traffic distribution to achieve multi-path packet transmission, solving the problem of load imbalance in traditional cross-data center links and helping to improve resource utilization in AI computing networks.

The industry's first verification of the end-to-end Scale-Across solution on a 800G platform not only demonstrates the forward-looking capabilities of high-end data communication equipment in ultra-large-scale computing networks, but also provides a replicable benchmark for building integrated computing networks and advancing intelligent and service-oriented computing services.

Moving forward, all parties will continue to promote the trial deployment of Scale-Across technology at national-level computing hubs and jointly conduct research on cross-vendor interoperability standards, accelerating the development of an open and decoupled intelligent computing ecosystem.

Source: ZTE and China Mobile media announcement

Follow @PipelineWire