Skip to main content

CX664D-N

64-Port 200G QSFP56 Low Latency Leaf Spine Switch for AI/ML/Data Center, Enterprise SONiC

  • Powered by Asterfusion’s Enterprise SONiC Distribution
  • Industry-leading ultra-low latency: 500ns with Teralynx 7
  • Support: RoCEv2, EVPN Multihoming, BGP-EVPN, INT, and more
  • Purpose-built for AI, ML, HPC, and Data Center applications
Please login to request a quote

200GbE Data Center Switch

Enterprise SONiC 64-Port 200G QSFP56 Data Center Leaf/Spine Switch Low Latency 

The Asterfusion CX664P-N is a high-density 64-port 200G QSFP56 data center switch engineered specifically for spine architectures. Powered by the blazing-fast Marvell Teralynx 7 ASIC, it delivers line-rate L2/L3 switching performance up to an impressive 12.8 Tbps, combined with ultra-low latency and unparalleled packet processing capabilities.
With latency slashed to a remarkable 500 ns, the CX664P-N is purpose-built for the most demanding, latency-sensitive workloads like AI, HPC, and machine learning—ensuring rock-solid, consistent high-throughput performance even during intense bursts of small-packet traffic.

Enterprise SONiC 64-Port 200G QSFP56 Dater Center Leaf/Spine Switch Low Latency 

The Asterfusion CX664P-N is a high-density 64-port 200G QSFP56 data center switch engineered specifically for spine architectures. Powered by the blazing-fast Marvell Teralynx 7 ASIC, it delivers line-rate L2/L3 switching performance up to an impressive 12.8 Tbps, combined with ultra-low latency and unparalleled packet processing capabilities.
With latency slashed to a remarkable 500 nanoseconds, the CX664P-N is purpose-built for the most demanding, latency-sensitive workloads like AI, HPC, and machine learning—ensuring rock-solid, consistent high-throughput performance even during intense bursts of small-packet traffic.

500ns

Ultra Low Latency

7600Mpps

Packet Forwarding Rate

70MB

Packet Buffer

KEY POINTS

Key Facts

500ns

Ultra Low Latency

7600Mpps

Packet Forwarding Rate

70MB

Packet Buffer

Optimized Software
for Advanced Workloads
AsterNOS

Pre-installed with Enterprise SONiC, it supports cutting-edge features including RoCEv2 and EVPN multihoming, pairing lightning-fast forwarding rates with massive packet buffers. The CX664P-N doesn’t just meet expectations — it shatters them, outperforming competitors across the board.

Optimized Software for Advanced Workloads – AsterNOS

Pre-installed with Enterprise SONiC, it supports cutting-edge features including RoCEv2 and EVPN multihoming, pairing lightning-fast forwarding rates with massive packet buffers. The CX664P-N doesn’t just meet expectations — it shatters them, outperforming competitors across the board.

Automated O&M

Python and Ansible, ZTP
In-band Network Telemetry
SPAN / ERSPAN Monitoring

High Reliability

ECMP/ Elastic Hash
MC-LAG
BFD for BGP and OSPF

Zero Pacekt

RoCEv2
PFC/ PFC Watchdog / ECN
QCN / DCQCN / DCTCP

Virtualization

VXLAN
BGP EVPN
EVPN Multihoming

asternos-bg

Automated O&M

Python and Ansible, ZTP
In-band Network Telemetry
SPAN / ERSPAN Monitoring

High Reliability

ECMP/ Elastic Hash
MC-LAG
BFD for BGP and OSPF

Zero Pacekt

RoCEv2
PFC/ PFC Watchdog / ECN
QCN / DCQCN / DCTCP

Virtualization

VXLAN
BGP EVPN
EVPN Multihoming

Specifications

Ports64x 200G QSFP56
2x 10G SFP+
Switch ChipMarvell Teralynx 7 IVM77700
CPUBroadwell DE D-1508Number of VLANs4,094
Switching Capacity12.8 TbpsMAC Address Entries44000
Packet Forwarding Rate7,600 MppsJumbo Frame9,216
SDRAMDDR4 16 GBARP Entries36000
SSD Memory128 GBIPv4/IPv6 RoutesIPv4: 96,000 Host + 72,000 Prefix
IPv6: 47,000 Host + 36,000 Prefix
Packet Buffer70 MBLatency~500ns
Hot-swappable AC Power Supplies2 (1+1 Redundancy)Max. Power Consumption<792W
Hot-swappable Fans4 (3+1 Redundancy)MTBF>100,000 Hours
Dimensions (HxWxD)88.8x 440x 560mmRack Units2 RU
Weight16kgThermal AirflowFront-to-Rear

Quality Certifications

ISO-9001-SGS
ISO-14001-SGS
ISO-45001-SGS
ISO-IEC-27001-SGS
fcc-icon
ce-icon

Features

Marvell Teralynx 7 Inside: Unleash Industry-Leading Performance & Ultra-Low Latency

With Marvell Teralynx 7 at its core, Asterfusion 200G switches
achieve blazing-fast sub-500ns latency—perfect for AI/ML, HPC, and NVMe where speed drives results.
marvell-teralynx7
12.8 Tbps

Switching Capacity

500ns

End-to-end Latency

70 MB

Buffer Size

12.8 Tbps

Switching Capacity

500ns

Ultra Low Latency

70 MB

Buffer
Size

Full RoCEv2 Support for Ultra-Low Latency and Lossless Networking

Asterfusion 200G switches deliver full RoCEv2 support with ultra-low latency and
near-zero CPU overhead. Combined with PFC and ECN for advanced congestion control, it enables a true lossless network.
rocev2-vs-tcp

Asterfusion Network Monitoring & Visualization

AsterNOS supports Node Exporter to send CPU, traffic, packet loss, latency, and RoCE congestion metrics to Prometheus.
Paired with Grafana, it enables real-time, visual insight into network performance.
prometheus-paired-with-grafana

O&M in Minutes, Not Days

Automate with Easy RoCE, ZTP, Python and Ansible,  SPAN / ERSPAN Monitoring & more—cutting config time, errors, and costs.
200G switch operation and management

Comparison of Data Center Switches

CX864E-NCX732Q-NCX664D-NCX564P-NCX532P-N
Port Speeds64x800G, 2x10G32x400G, 2x10G64x200G, 2x10G64x100G, 2x10G32x100G, 2x10G
Hot-swappable PSUs and FANs1+1 PSUs, 3+1 FANs1+1 PSUs, 5+1 FANs1+1 PSUs, 3+1 FANs1+1PSUs, 3+1 FANs1+1 PSUs, 4+1 FANs
MC-LAG
BGP/MP-BGP
EVPN
BGP EVPN-Multihoming
RoCEv2
PFC
ECN
In-band Telemetry
Packet Spray××××
Flowlet××××
Auto Load Balancing××××
INT-based Routing××××
WCMP××××

Panel

What’s in box

FAQs

Does the switch support RoCEv2? What specific ECN, PFC, and DCB features are supported?
Supports RoCEv2, ECN/PFC/DCBX, and Easy RoCE, and configures the above parameters with one click.
What is the low-latency performance of Asterfusion’s 64×200G switch? What latency-sensitive applications is it suitable for?
Our CX664D-N switch has a low latency of 500ns, making it suitable for latency-sensitive applications such as AI inference, HPC, and edge computing.
Does the switch support containerized applications (e.g., PB, INT, ZTP)? Is it based on the FRR routing stack?
Our AsterNOS inherits SONiC’s containerized architecture, fully supporting containerized applications such as Packet Broker and In-band Network Telemetry. Additionally, based on the FRRouting (FRR) routing stack, it provides robust and reliable control plane functionality.
Which RDMA-capable NICs (e.g., Mellanox ConnectX-4/5/6, Broadcom, Intel) and drivers have been validated on your CX-N series switches?

All major RDMA-capable NICs, including Mellanox ConnectX-4/5/6, Broadcom, and Intel, as well as their standard drivers, have been validated on our CX-N series switches.

If you have any recommendations for setting up an NVMe-oF, any documentation/case studies?

We have successfully deployed NVMe-oF solutions for several customers. Please contact bd@cloudswit.ch to get case study~

We want your switches to do AI inference, Our working AI inference config looks somewhat like: Camera (SRT) -> Compute Node NIC -> GPU -> [Compute Node NIC -> NVMe-oF + Compute Node NIC -> Other Nodes for processing]. If you have any past setups similar to that, are there any solutions to include Camera SRT feed or GigE feed into RoCEv2 congestion management?

Yes, by marking or prioritizing the camera SRT or GigE traffic at the switch egress or ingress, the switches can treat it as high-priority and apply PFC/ECN-based congestion management, similar to RDMA traffic. However, SRT itself implements its end-to-end congestion control at the application layer and does not utilize or require DCQCN, which is specific to RDMA flows.