The Cisco Nexus 3408-S is a 4RU, 8-slot data center switch designed for high-density 100G and 400G deployments. Its 12.8-Tbps ASIC architecture supports configurations of up to 128 native 100G ports or 32 native 400G ports, with breakout options for lower-speed connectivity. The platform provides a 25.6-Tbps switching capacity, wire-rate Layer 2 and Layer 3 forwarding, deep buffer monitoring, programmable forwarding profiles, and modular NX-OS operations.
Hardware architecture and port density
The Nexus 3408-S uses Line-Card Expansion Modules, or LEMs, installed in an eight-slot chassis. Two LEM types are identified:
| LEM | Port type | Native speeds | Breakout capability | Best For |
|---|---|---|---|---|
| NXM-X16C | 16 x QSFP28 | 100G | 2 x 50G and 4 x 25G/10G | High-density 100G leaf, spine, and aggregation deployments |
| NXM-X4D | 4 x QSFP-DD | 400G, 200G, 100G, 40G | 2 x 200G/50G or 4 x 100G/50G/25G/10G | 400G spine, fabric interconnect, and high-bandwidth data center designs |
The NXM-X16C provides 16 QSFP28 ports per module. Across eight occupied slots, the chassis can provide up to 128 native 100G ports. The NXM-X4D provides four QSFP-DD ports per module. Across eight occupied slots, the chassis can provide up to 32 native 400G ports.
The platform also supports broader breakout combinations. The chassis supports breakout for 2 x 200G/50G, 4 x 100G/50G/25G/10G, and 8 x 50G, with up to 128 ports of 100G or up to 168 ports of 50G. Breakout support is dependent on the specific LEM and software configuration. The NXM-X16C supports 2 x 50G and 4 x 25G/10G breakout. The NXM-X4D supports 2 x 200G/50G and 4 x 100G/50G/25G/10G breakout.
QSFP-DD ports are backward-compatible with QSFP+, QSFP28, and QSFP56. Each QSFP-DD port can operate at 400G, 200G, 100G, or 40G native speed, and supports breakout at 200G, 100G, 50G, 25G, and 10G.
Chassis components
The base chassis includes the following components:
- Eight LEM slots
- Dual-redundant power supplies
- Redundant fan infrastructure
- Beacon LED
- Status LED
- One RS-232 serial console port
- One RJ45 management port
- One SFP management port
- Two 100/1000 Mbps SFP ports
- One USB port
The chassis supports both HVAC and DC power inputs according to the product overview and ordering information. The hardware specification table identifies AC forward-airflow operation, while the ordering table lists high-voltage AC, DC, and AC power supply options. Power selection must therefore be aligned with the intended facility feed and the required airflow direction.
Switching performance and scalability
The platform provides 25.6 Tbps of switching capacity. The forwarding architecture is described as wire-rate Layer 2 and Layer 3 switching on all ports, with up to 25.6 Tbps and 7.2 billion packets per second at ingress and 10 billion packets per second at egress. The wire-rate qualification applies to packets greater than 200 bytes.
The switch includes a four-core CPU, 32 GB of DRAM, 70 MB of dynamic buffer allocation, and 128 GB of boot flash memory. These resources support large Layer 2 and Layer 3 tables, programmability, packet analysis, telemetry-related functions, and high-density data center operations.
The hardware scalability values listed for normal mode are:
| Resource | Scale |
|---|---|
| MAC addresses | 120K |
| IPv4 unicast routes | 440K |
| IPv6 unicast routes | 360K |
| IPv4 hosts | 192K |
| IPv6 hosts | 96K |
| IPv4 multicast routes | Up to 96K with 8K groups |
| VLANs | 4K |
| ACL entries | 3.5K ingress and 1.5K egress |
| RSTP instances | 123 |
| MST instances | 64 |
| EtherChannels | 512 |
| Ports per EtherChannel | Up to 128 |
| Buffer size | 70 MB |
| Boot flash | 128 GB |
The datasheet also lists 1024 total Layer 3 interfaces, including routed ports, SVIs, PortChannels, and subinterfaces. The general software feature table identifies 64-way ECMP, while the platform overview identifies 512-way ECMP as an advanced scalability capability. Engineering designs should validate the intended topology and software release against the applicable verified scalability guide.
Layer 2 and Layer 3 functions
Layer 2 functionality includes VLAN switching, IEEE 802.1Q trunking, LACP, UDLD in standard and aggressive modes, RSTP, MSTP, STP guard features, PortFast, Root Guard, Bridge Assurance, and vPC.
The platform supports up to 4096 VLANs and jumbo frames up to 9216 bytes on all ports. EtherChannel hashing can use Layer 2, Layer 3, and Layer 4 information. Link-level flow control is supported through IEEE 802.3x.
Layer 3 capabilities include:
- Routed interfaces
- Switch Virtual Interfaces
- PortChannels
- Subinterfaces
- Static routing
- RIPv2
- EIGRP
- OSPF
- BGP
- HSRP
- VRRP
- 64-way ECMP in the general Layer 3 feature table
- VRF-Lite
- VRF-aware unicast
- VRF-aware multicast
- VRF route leaking
- Routed Layer 3 and Layer 4 ACLs
- Jumbo frames up to 9216 bytes
The base license supports inter-VLAN routing, static routes, RIPv2, OSPFv2 with a stated limit of 256 routes, EIGRP stub, HSRP, and VRRP. The Essential License expands the routing set with full OSPFv2, EIGRP, BGP, IS-IS, and VRF-Lite, and adds DCNM and telemetry functions.
High-availability design
The chassis includes dual-redundant power supplies and hot-swappable power supplies and fans. The overview describes redundant 2+1 fans, while the hardware table identifies six individual hot-swappable fans in a 5+1 redundant arrangement. The latter provides the detailed hardware configuration.
vPC provides Layer 2 multipathing and reduces reliance on Spanning Tree Protocol for active-active designs. ECMP supports Layer 3 fat-tree architectures and provides multiple forwarding paths for resiliency and capacity expansion. Advanced reboot capabilities include hot and cold patching.
Presales designs should account for the following:
- Use two independent power feeds when the facility design supports them.
- Confirm the required airflow direction before ordering power supplies and fan assemblies.
- Validate LEM population against the required native port count and breakout plan.
- Use vPC for Layer 2 dual-homing requirements and ECMP for routed fabric designs.
- Confirm software-specific scalability for ECMP, routing tables, breakout modes, and advanced reboot functions.
- Reserve chassis slots for future capacity only after verifying the expected power, cooling, and port-density impact.
NX-OS operations and visibility
Cisco NX-OS provides a modular operating system with CLI, XML, NX-API, Python, Linux containers, and JSON API support. The platform supports Power-On Auto Provisioning, Python scripting, Embedded Event Manager, configuration rollback, and remote operational workflows.
Monitoring and troubleshooting features include:
- Advanced buffer monitoring
- Per-port and per-queue buffer utilization
- SPAN
- ERSPAN Versions 2 and 3
- Ingress and egress packet counters
- Ethanalyzer embedded packet analysis
- Comprehensive bootup diagnostics
- Cisco Online Health Management System
- Active buffer monitoring
- Syslog
- SNMP Versions 1, 2, and 3
- RMON
- Cisco Discovery Protocol
- NTP
The platform supports telemetry-related functions under the Essential License, including Buffer Drop Capture, High Delay Capture, and INT. These functions are relevant for environments where queue behavior, congestion, packet drops, and latency need to be correlated with application traffic.
Management security includes AAA, RBAC, RADIUS, TACACS+, SSHv2, AES for management traffic, digital certificates for management between the switch and RADIUS server, CoPP, VTY ACLs, and unified username and password support across CLI and SNMP. Telnet is also listed as supported, but secure management designs should use SSHv2.
Licensing model
| Software package | Main functions | Best For |
|---|---|---|
| System default | Layer 2 switching, VLANs, IEEE 802.1Q, LACP, UDLD, MSTP, RSTP, STP guard, AAA, ACLs, storm control, CoPP, DCNM support, SSHv2, SNMP, syslog, advanced buffer monitoring, SPAN, and ERSPAN | Layer 2 fabrics and basic operational deployments |
| Base license | Inter-VLAN routing, static routing, RIPv2, ACLs, OSPFv2 limited to 256 routes, EIGRP stub, HSRP, and VRRP | Basic Layer 3 access, aggregation, or small routed environments |
| Essential License, N3K-ES-XM | Advanced OSPFv2, EIGRP, BGP, IS-IS, VRF-Lite, DCNM, Buffer Drop Capture, High Delay Capture, and INT | Enterprise Layer 3 fabrics, BGP/IS-IS designs, multi-tenant routing, and advanced telemetry |
| Cisco Nexus Data Broker license, NDB-FX-SWT-K9 | Tap and SPAN aggregation functions, supported on the Essential License | Centralized monitoring tool aggregation and traffic distribution |
A presales bill of materials should include the Essential License when the design requires BGP, IS-IS, full OSPFv2 capability, VRF-Lite, or the listed telemetry features. Cisco Nexus Data Broker functions require both the Data Broker license and the Essential License.
Optics, cabling, and breakout planning
Supported connectivity includes fiber and copper cabling for 10G, 25G, 40G, 50G, 100G, and 400G. Active Optical Cables and Direct-Attached Cables are supported. Optical module selection must be matched to the LEM type, native speed, breakout mode, reach requirement, and required software release.
The 100G NXM-X16C is based on QSFP28. The 400G NXM-X4D is based on QSFP-DD and supports backward compatibility with QSFP+, QSFP28, and QSFP56. A port-count calculation should distinguish between:
- Native port count
- Logical breakout port count
- Transceiver lane requirements
- Cable type and reach
- Host-facing speed
- Fabric-facing speed
A 400G port used as four 100G breakout interfaces is not equivalent to four independent native 400G ports. Presales documentation should identify both the physical port allocation and the resulting logical interface count.
Power, airflow, and PoE considerations
The chassis uses two redundant power supplies. Listed power options include:
- NXA-PHV-2KW-PI: 2KW high-voltage power supply, forward airflow
- NXA-PDC-2KW-PI: 2KW DC power supply, forward airflow
- NXA-PAC-2KW-PI: 2KW AC power supply, forward airflow
The hardware table specifies 100 to 240 VAC input, 50 to 60 Hz frequency, and 89 to 91 percent power supply efficiency at 220V. The cooling design is forward airflow, with air entering through the port side and exiting through the fan tray and power supplies.
No PoE support or PoE power budget is specified. The Nexus 3408-S should therefore be sized as a data center switching platform rather than as a powered access switch. Endpoints requiring PoE need a separate PoE-capable access layer or power solution.
Environmental and physical requirements
The chassis dimensions are 6.97 inches high by 17.3 inches wide by 31.6 inches deep. This is a 4RU platform with significant depth and weight, so rack depth, rear clearance, cable bend radius, lift access, and installation logistics must be checked before order placement.
Published weight values are configuration-dependent:
| Configuration | Weight |
|---|---|
| Chassis without LEMs, power supplies, or fans | 68 lb |
| With eight 100G LEMs, power supplies, and fans | 107.7 lb |
| With eight 400G LEMs, power supplies, and fans | 104.5 lb |
| NXM-X16C individual LEM | 3.6 lb |
| NXM-X4D individual LEM | 3.2 lb |
Environmental limits are:
- Operating temperature: 32 to 104 degrees F, or 0 to 40 degrees C
- Storage temperature: -40 to 158 degrees F, or -40 to 70 degrees C
- Relative humidity: 5 to 95 percent, non-condensing
- Operating altitude: Up to 13,123 feet
- Non-operating altitude: Up to 16,000 feet
MTBF is not specified in the supplied product specifications. Acoustic noise is also not specified. These values should be obtained from the applicable hardware installation or environmental documentation when required for facility design, colocation approval, or acoustic modeling.
The platform is listed against safety, EMC emissions, EMC immunity, and RoHS requirements. Compliance references include UL 60950-1, CAN/CSA-C22.2 No. 60950-1, EN 60950-1, IEC 60950-1, AS/NZS 60950-1, GB4943, Class A emissions standards, EN55024, CISPR24, EN300386, KN24, and RoHS 5 compliance except for lead press-fit connectors.
Ordering matrix
| Part number | Description | Best For |
|---|---|---|
| NXM-X16C | Nexus 100G Line Expansion Module | 100G port density and QSFP28-based access or fabric connectivity |
| NXM-X4D | Nexus 400G Line Expansion Module | 400G spine and high-bandwidth interconnects |
| NXM-XBLNK | Nexus blank Line Expansion Module | Slot blanking and unused-slot management |
| NXA-FAN-300CFM-PI | Forward-airflow Nexus fan | Replacement or maintenance stock for the fan assembly |
| NXA-PHV-2KW-PI | 2KW high-voltage power supply, forward airflow | High-voltage facility power with port-side intake |
| NXA-PDC-2KW-PI | 2KW DC power supply, forward airflow | DC-powered data center deployments |
| NXA-PAC-2KW-PI | 2KW AC power supply, forward airflow | AC-powered data center deployments |
| N3K-ES-XM | Essential License with Layer 3 LAN Enterprise, DCNM, and telemetry features | Advanced routing, VRF-Lite, BGP, IS-IS, and telemetry |
| N3K-C3408-S= | Nexus 3408-S switch with 32 QSFP-DD ports, spare | Complete chassis replacement or spare inventory |
| NXM-X16C= | 100G LEM spare | Field replacement for an NXM-X16C |
| NXM-X4D= | 400G LEM spare | Field replacement for an NXM-X4D |
| NXM-XBLNK= | Blank LEM spare | Replacement slot blank |
| NXA-FAN-300CFM-PI= | Forward-airflow fan spare | Fan replacement stock |
| NXA-PHV-2KW-PI= | High-voltage 2KW power supply spare | High-voltage PSU replacement stock |
| NXA-PDC-2KW-PI= | DC 2KW power supply spare | DC PSU replacement stock |
| NXA-PAC-2KW-PI= | AC 2KW power supply spare | AC PSU replacement stock |
| NXB-CPU-FRU= | CPU module spare | Field replacement of the CPU module |
Warranty and service
The Nexus 3000 Series hardware warranty is a one-year limited hardware warranty. Hardware replacement is provided with a ten-day turnaround from receipt of a Return Materials Authorization.
Available service options include Cisco Advanced Services and Cisco SMARTnet Service. Advanced Services uses an architecture-led approach for data center infrastructure planning, deployment, optimization, migration, and operational improvement. SMARTnet Service provides access to Cisco network experts and support resources for mission-critical issues. Smart Call Home capabilities provide proactive diagnostics and real-time alerts for supported Nexus 3000 Series switches.
Service selection should reflect the operational role of the chassis, replacement urgency, spare strategy, and required access to expert support. For high-density fabric roles, customers should also define an internal spare policy for power supplies, fans, LEMs, and the CPU module.