Data center networks are evolving in more than one direction. Spine and core links may move to 400G or 800G, while many server-facing connections remain at 10G, 25G, 50G, or 100G. That leaves a clear design challenge at the leaf: add more 100G capacity while making efficient use of rack space. AI, high-performance computing (HPC), and hyperscale cloud environments are increasing demand for high-throughput connections. For many deployments, 100G remains a cost-efficient workhorse, and the move to faster spine links can increase the need for dense 100G leaf capacity.
The Cisco N9348C8D-H2 addresses this need by combining dense, flexible 100G access with higher-speed uplinks in a compact 1RU design.
Dense 100G in a compact design
The Cisco N9348C8D-H2 is a fixed 1RU switch that brings together 48 100G SFP-DD ports and eight 400G QSFP-DD uplinks. SFP-DD ports support 10G, 25G, 50G, and 100G; QSFP-DD uplinks support 40G, 100G, 200G, and 400G. Together, these options let organizations connect varied endpoints at the leaf and link to faster spine layers as the fabric grows.
Built on the Cisco Cloud Scale LS12800 H2 12.8T ASIC, the switch uses 256 × 50GE SerDes and delivers up to 8 Tbps of forwarding capacity. It supports up to 2 million routes, 14,000 shared ingress/egress classification TCAM entries, and an 80 MB shared on-die packet buffer.
Figure 2. Front and rear views of the Cisco N9348C8D-H2
Why SFP-DD matters
SFP-DD is a compact, double-density interface that carries 100G over two 56G PAM4 electrical lanes, enabling this port density. Support for 10G, 25G, and 50G also gives customers a gradual path to 100G while helping data centers make efficient use of rack space as traffic grows.
The switch supports SFP-DD direct-attach cables and optical transceivers. With supported connections, SFP-DD ports can link to SFP-DD or QSFP28 endpoints or break out to two 50G SFP56 endpoints; they also support compatible 100G network interface cards (NICs). QSFP-DD uplinks can connect to 400G endpoints or break out to four 100G QSFP28 endpoints, as shown.
Built for choice across the network
Cisco Nexus One provides a broader data center networking architecture for choice across network operations, bringing together silicon, systems, optics, software, and operations. Within this architecture, the Cisco N9348C8D-H2 supports Cisco NX-OS or ACI on the same hardware, with Cisco Nexus Dashboard or Application Policy Infrastructure Controller (APIC) management in the corresponding environment.
For enterprise data centers and financial firms adding 100G capacity within existing rack space, the Cisco N9348C8D-H2 offers a compact leaf platform. Customers can deploy it in a new 100G fabric or transition from 10G, 25G, or 50G access to 100G. Eight 400G uplinks connect the leaf to faster spine layers.
Designed for data center operations
To support day-to-day operations, Cisco Cloud Scale telemetry includes flow telemetry (FT), flow table event (FTE), and streaming statistics export (SSX) capabilities for monitoring and troubleshooting. Precision timing protocol (PTP) supports timing requirements. Front-to-rear or rear-to-front airflow and AC or DC power options help the switch fit different rack and facility designs.
Bring dense 100G into your fabric
Together, dense 100G access, 400G uplinks, software choice, and hardware visibility help enterprise and financial services organizations expand leaf capacity seamlessly. The Cisco N9348C8D-H2 connects current workloads to a fabric that can evolve with them.
Ready to plan your next data center fabric? Join us at Cisco Live Melbourne, November 9-12, to see the Cisco N9348C8D-H2 in action through demos and talk with our experts about building a data center fabric that can evolve with your needs.
Additional resources:
