CloudEngine 16800 includes a 48-port 400 GE line card per slot, and is able to evolve to the autonomous driving network, anticipating future digital traffic demand. It provides powerful heat dissipation through SuperCooling, reducing the power consumption by 50% per data bit.
Embedded with an AI chip, CloudEngine 16800 uses the unique iLossless algorithm to learn and train network-wide traffic in real time, implementing network adaptation and self-optimization. With zero packet loss and E2E μs latency, it achieves maximum throughput.
CloudEngine 16800 provides Telemetry-based data collection in milliseconds, supporting the intelligent O&M platform. Its embedded AI chip supports edge AI processing and local inference and fault management.
|Model||CloudEngine 16804||CloudEngine 16808||CloudEngine 16816|
|Switching Capacity||43/387 Tbit/s2||86/774 Tbit/s2||173/1548 Tbit/s2|
|Forwarding Performance||11,280 Mpps||22,560 Mpps||45,120 Mpps|
|Switching Fabric Module Slots||6 (scalable to 9 for future expansion)|
|Fabric Architecture||Clos architecture, cell switching, VoQ|
|Airflow Design||Strict front-to-back airflow design|
VS (1:16 virtualization)
Cluster Switch System (CSS)3
VXLAN and VXLAN bridging
QinQ access VXLAN
|L2/L3||VLAN, STP, LACP
Static route, IPv4/IPv6 dynamic route protocol
IP packet fragmentation and reassembly
Hardware-based Bidirectional Forwarding Detection (BFD)
|Programmability||Standard NETCONF interface
Ansible-based automatic configuration and open-source module release
1 This content is applicable only to regions outside mainland China. Huawei reserves the right to interpret this content.
3 For details about the configuration, please see: http://support.huawei.com/onlinetoolsweb/virtual/en/dc/stack_index.html?dcf.