Inspur I24NM-4S 25GbE Top-of-Rack Switch

The Inspur I24NM-4S is a cutting-edge 48-port 25GbE plus 8-port 100GbE top-of-rack (ToR) switch engineered for modern hyperscale and high-performance computing environments. Running open network operating systems like SONiC, the switch delivers disaggregated networking capabilities that decouple hardware from software, enabling operators to standardize on open-source NOS platforms while leveraging merchant silicon architectures. With 1.6 Tbps aggregate throughput and support for high-density port configurations, the I24NM-4S addresses the evolving demands of cloud builders and HPC operators seeking cost-effective, vendor-neutral switching solutions.

The switch architecture emphasizes operational flexibility through open networking standards. By supporting SONiC and compatible open NOS platforms, the I24NM-4S eliminates vendor lock-in while enabling rapid feature deployment and customized network policies tailored to specific workload requirements. The combination of 25GbE server connectivity and 100GbE uplink capacity creates flexible oversubscription ratios suitable for both east-west and north-south traffic patterns in modern data centers. Hardware features include advanced packet buffering, non-blocking fabric design, and support for deep packet inspection and flow-based networking.

Launched in Q2 2023, the I24NM-4S reflects Inspur's commitment to the disaggregated networking movement, directly addressing the infrastructure preferences of tier-1 cloud operators and research institutions transitioning to open-source network control planes. The platform supports standards-based protocols including OpenFlow, VXLAN, and BGP unnumbered configurations, making it compatible with existing ecosystem tools and operational workflows.

Ideal for

• Hyperscale cloud data center leaf-spine fabric deployments requiring standardized merchant silicon switching with open NOS flexibility
• High-performance computing (HPC) clusters with dense east-west traffic patterns between compute nodes and storage infrastructure
• Disaggregated networking deployments where operators maintain multiple open-source NOS instances across heterogeneous switch hardware
• Research and academic institutions building software-defined networking labs and production environments with cost-optimized switching hardware
• Multi-tenant cloud environments requiring granular traffic engineering, quality-of-service policies, and tenant isolation via open programmable data planes
• AI/ML training infrastructure with high-bandwidth requirements between GPU clusters and distributed storage systems

Technical specifications

Available from Omnixon Global. Submit an RFQ and our team will confirm configuration and availability for your order.