● Model No.: CE9855-32DQ
● Port details: 32 x 400GE QSFP-DD + 2 x 10GE SFP+
● Application scenarios: 400G access/aggregation layer for hyperscale cloud data centers, AI computing clusters, and distributed storage.
● Feature: It has 32 400GE QSFP-DD ports, with a switching capacity of 25.6Tbps and a packet forwarding rate of up to 8000Mpps.
● Warranty: 1years
● MOQ: 1 pieces
Application scenarios
400G Access/Aggregation Layer for Ultra-Large Cloud Data Centers
Serving as a 400G access switch, it connects to 100G/25G TOR switches, supporting large-scale cloud services and east-west traffic.
AI Computing Clusters/GPU Clusters/Large Model Training
Ultra-high bandwidth, low latency, and lossless transmission meet the needs of high-speed interconnection of GPUs/DPUs and massive data interaction.
High-Performance Computing (HPC) and Supercomputing Centers
Non-blocking and low latency, suitable for supercomputing scenarios such as scientific computing and simulation.
Distributed Storage and Big Data Platform Backbone
High throughput and large cache ensure the stability of massive data read/write, replication, and analysis.
Finance, Securities, and Government/Enterprise Core Networks
High reliability, low latency, and high security serve as aggregation nodes for financial transactions and core government/enterprise businesses.
Key selling points
1. 400G Ultra-High Density, Flexible Splitting
Integrates 32x400GE QSFP-DD ports in a 2U space, with a maximum support of 128x100GE per device, meeting the ultra-high bandwidth requirements of AI clusters, supercomputing, and distributed storage.
The 400GE ports can be split into 4x100GE or 2x200GE, flexibly adapting to access and aggregation layer networking.
2. Non-blocking Full Line-Speed, Low Latency, Large Buffer
Non-blocking switching architecture ensures line-speed forwarding on all ports, eliminating bandwidth bottlenecks.
Hardware-level low-latency forwarding + 65MB large buffer adapts to high-frequency transactions, real-time computing, and RoCEv2 lossless traffic.
3. iLossless Intelligent Lossless, Adaptable to AI/Storage
Supports the iLossless intelligent lossless algorithm, providing a "no packet loss, low latency, high throughput" network environment for RDMA/RoCEv2 traffic.
Supports lossless data center features such as PFC, ECN, and DCQCN, ensuring stable operation of AI training and distributed storage.
4. Cloud Computing Full-Scenario Features
Supports VXLAN, EVPN, SRv6, SDN, Telemetry, and INT in-path detection.
Improves IPv6, multicast, security, QoS, and traffic visualization, adapting to cloud-native and intelligent computing networks.
5. High Reliability, Easy Operation and Maintenance, Green Energy Saving
Dual power supplies + dual fan redundancy, supporting hot-swapping, BFD, and ISSU non-disruptive upgrades.
Supports M-LAG, stacking, and cross-device link aggregation, enhancing network resilience.
Intelligent fan speed control and low-power chips create a green and low-carbon data center.
Detailed description
| parameter item | Specifications |
| model | CE9855-32DQ |
| Device form factor | 2U fixed box data center switch (excluding power fan) |
| business port | 32 × 400GE QSFP-DD+2 × 10GE SFP+(no distinction between upstream and downstream) |
| port rate | 400GE/200GE/100GE/40GE adaptive; 400GE detachable 4 × 100GE/2 × 200GE |
| switching capacity | 25.6 Tbps |
| Packet forwarding rate | 8000 Mpps |
| cache | 65 MB |
| Routing table entry | 2M IPv4 / 1M IPv6 |
| virtualization | M-LAG、iStack、EVPN、VXLAN、SRv6 |
| power supply | Support 2 redundant pluggable power supplies (AC/DC/HVDC) |
| fan | Dual redundant pluggable fans |
| air duct | Port side air intake (front to back)/reverse air duct optional |
| size | 442 × 600 × 43 mm(1U) |
| Maximum Power Consumption | Approximately 528 W |
| operating system | VRP V8 (V300R024C00 and above) |
Email: Lilicheng0510@163.com