Cisco N3K-C31128PQ-10GE Full Product Description
1. Short Overview
N3K-C31128PQ-10GEis a high-density 2RU second-generation low-latency Layer 2/3 data center leaf/ToR switch from the Cisco Nexus 3100 series, released June 8, 2015, End-of-Sale (Feb 25, 2022) and End-of-Support-Life (Feb 28, 2027)Cisco. It is a bare chassis SKU without factory pre-installed power supplies, fan trays and optical transceivers, designed for high-density virtualization racks, HPC, cloud multi-tenant leaf-spine fabrics, and unified LAN/FCoE storage convergence.
This all-fiber switch delivers96 auto-sensing SFP+ 1/10G downlink portsplus8 universal 40G QSFP+ uplink ports. Each QSFP+ slot supports native 40GE or breakout to four independent 10GE channels via QSFP-to-4×SFP+ splitter transceivers. Running modular Cisco NX-OS, it provides full line-rate non-blocking forwarding, ultra-low deterministic microsecond latency, dual airflow FRU options (port-side intake / port-side exhaust), carrier-grade redundant hot-swappable AC/DC power and cooling FRUs, a 12 MB shared packet buffer, 64 GB SSD boot media, and programmable SDN capabilities including VXLAN and OpenFlow.
-
Suffix breakdown definition:
-
PQ= 96 × SFP+ fiber downlinks + 8 universal 40G QSFP+ fabric uplink slots
-
-10GE= Base bare chassis optimized for 10G fiber server access deployments; no bundled power, fan or optical FRUs included
-
Core advantages: Industry-leading port density within a compact 2RU rack footprint, PHY-less low-latency silicon design, flexible dual airflow compatibility to match hot/cold aisle rack standards, native interoperability with Nexus 2000/2300 blade & rack FEX devices, POAP zero-touch provisioning, and single-pane unified fabric management via Cisco DCNM.
2. Part Number Suffix Breakdown & Cross-Model Comparison
Full PN Breakdown
-
N3K: Cisco Nexus 3000 Series second-generation low-latency programmable data center Layer 2/3 switch family
-
C31128PQ-10GE: 2RU high-density all-fiber leaf switch bare chassis
-
31128: High-density aggregate port platform with total 104 combined downlink + uplink interfaces
-
PQ: 96 × multi-speed SFP+ fiber host ports + eight universal 40G QSFP+ fabric uplink ports
-
10GE: Bare unpopulated chassis variant optimized for 10G fiber access workloads; no factory pre-installed power supplies, fan trays or optical transceivers
-
Airflow flexibility: Two separate FRU families available: Port-side Intake (PI / reverse airflow) and Port-side Exhaust (PE / forward airflow). Mismatched airflow power supply and fan tray modules trigger thermal protection shutdown.
Hardware Comparison Table
|
Item
|
N3K-C31128PQ-10GE (96×10G SFP+ 2RU Bare Chassis)
|
N3K-C31108PC-V (48×10G SFP+ 1RU Leaf)
|
N3K-C31108TC-V (48×10GBASE-T Copper ToR)
|
|
Form Factor
|
2RU 19-inch rack mount
|
1RU compact leaf
|
1RU compact copper ToR
|
|
Downlink Ports
|
96 × auto-sense 1G/10G SFP+ Fiber
|
48 × auto-sense 1G/10G SFP+ Fiber
|
48 × 10GBASE-T RJ45 Copper
|
|
Uplink Interfaces
|
8 × 40G QSFP+ (breakout to 4×10G each)
|
6 × QSFP28 (100G/40G/10G breakout)
|
6 × QSFP28
|
|
Switching Capacity
|
2.5 Tbps non-blocking full-duplex
|
2.8 Tbps
|
2.8 Tbps
|
|
Shared Buffer
|
12 MB unified dynamic buffer
|
16 MB deep buffer
|
16 MB deep buffer
|
|
Storage
|
64 GB onboard SSD boot media
|
64 GB SSD
|
64 GB SSD
|
|
Native VXLAN
|
Software-supported VXLAN bridging
|
Hardware-offloaded VXLAN/EVPN
|
Hardware-offloaded VXLAN/EVPN
|
|
Airflow Options
|
Dual PI / PE airflow FRUs selectable
|
Fixed PI airflow pre-populated
|
Fixed PI airflow pre-populated
|
|
Pre-installed FRUs/Optics
|
None (pure bare chassis)
|
Dual PI PSUs + fan tray integrated
|
Dual PI PSUs + fan tray integrated
|
|
Oversubscription Ratio
|
Fixed 2:1 (960G down / 320G uplink)
|
Fixed 2:1
|
Fixed 2:1
|
|
Primary Use Case
|
High-density all-fiber virtualization rack leaf/spine
|
Modern 100G uplink cloud leaf deployments
|
Copper-based 10G server rack access
|
Factory Bare Chassis Included Hardware
-
2RU standard 19-inch horizontal rack-mount passive switch chassis
-
Rear port panel interface bank:
-
96 × auto-sensing 1G/10G SFP+ multi-speed fiber downlink ports for blade servers, virtualization hosts, IP SAN storage appliances
-
8 × universal 40G QSFP+ fabric uplink ports supporting SR4/LR4/ER4/BiDi optical transceivers, passive Twinax DAC, active AOC fiber cables, and QSFP-to-4×SFP+ breakout splitter modules
-
Front integrated out-of-band management subsystem:
-
RJ45 hybrid RS232 serial console port
-
Dedicated 10/100/1000BASE-T out-of-band MGMT Ethernet port
-
Dual USB Type-A flash storage ports for offline firmware upload, configuration backup and POAP provisioning files
-
Full front-panel LED status array: Global system health (green normal / amber fault), individual PSU fault alarms, fan tray fault indicators, per-port dual-color link/activity LEDs, blue UID locator beacon for rack asset identification during maintenance
-
Dual hot-swappable power supply bays (compatible with separate PI / PE airflow AC/DC power FRUs)
-
Single hot-swappable fan tray slot with multiple redundant internal fans; airflow direction determined by matching fan/PSU FRU selection
-
Standard 19-inch rack mounting rails, front cable management brackets, hardware installation guide and global safety & regulatory compliance documentation
-
No power supplies, fan trays or optical transceivers pre-installed (all FRUs sold separately as spare parts)
Items Not Included (Must Be Ordered Separately)
-
Hot-swappable AC/DC redundant power supplies (500W / 650W rating, port-intake PI or port-exhaust PE airflow variants optional; dual PSUs recommended for 1+1 power redundancy)
-
Single hot-swappable fan tray assembly with redundant internal fans; airflow direction must fully match the selected power supply modules
-
SFP+ transceivers for all 96 downlink ports: 1G/10G SR/LR/ER optical modules, passive Twinax DAC, active AOC fiber cables
-
40G QSFP+ transceivers for uplink ports: SR4, LR4, ER4, BiDi optics, 40G-to-4×10G breakout splitter modules, Twinax/AOC short-range cables
-
Optional accessories: Serial console rollover cable, out-of-band management Ethernet patch cords, lockable front door, air filter panel
-
Software licenses: N3K-BAS1K9 base license unlocks full Layer 3 routing functionality; base Layer 2 switching enabled without additional license fees; VXLAN bridging supported with standard NX-OS firmware upgrade
Supported Fabric & Parent Switch Compatibility
-
Full unified fabric + lossless DCB FCoE: Cisco Nexus 5500 / 5600 / 6000 Series (minimum NX-OS 7.0(3)I3(1))
-
Layer 2/Layer3 Ethernet only: Nexus 7000 / 7700 / 9000 Series
-
Native interoperability with Nexus 2000 blade & 2300 rack FEX devices for end-to-end consolidated LAN/SAN data center architecture
-
SDN compatibility: OpenFlow 1.3, NX-API REST APIs, Ansible/Puppet/Chef DevOps automation integration
3. Hardware Port & Slot Layout
Rear Panel 96 SFP+ Downlink & 8 QSFP+ Uplink Ports
-
96 × Auto-Sensing 1G/10G SFP+ Fiber Host Ports
IEEE 802.3ab/ae compliant multi-speed fiber interfaces with auto-negotiation of 1G or 10G line rates, supporting short-reach SR and long-reach LR/ER optical transceivers for cross-rack replication links. Implements IEEE DCB PFC/ETS lossless Ethernet standards for FCoE unified storage workloads, jumbo frames configurable up to 9216-byte MTU. All Layer 2 switching, ACL, MAC security and QoS policies are enforced locally on the switch hardware.
-
8 × Universal 40G QSFP+ High-Speed Fabric Uplink Ports
Two configurable operation modes for leaf-spine inter-switch connections:
-
Native 40GE mode: Direct high-bandwidth 40G inter-switch uplinks for scalable leaf-spine fabric core interconnections
-
Breakout mode: With QSFP-to-4×SFP+ splitter transceivers, each port expands to four independent 10GE fabric channels to multiply uplink port density and reduce optical hardware procurement costs
All eight uplink ports can be bundled into LACP fabric port-channels to support vPC / Enhanced vPC (EvPC) high-availability topologies, eliminating STP blocked links and maximizing uplink bandwidth utilization.
Front Panel Redundant FRU Slots & Out-of-Band Management Interfaces
-
Dual hot-swappable AC/DC power supply bays (supports 1+1 redundant power inputs for non-disruptive live PSU replacement without full switch reboot)
-
Single hot-swappable fan tray assembly with multiple redundant internal fans; airflow direction fully determined by matching fan and power supply FRU selection
-
Dedicated independent out-of-band management block: Dual USB flash ports, serial console port and 1G MGMT Ethernet port for offline firmware upgrade and configuration management, fully isolated from production data plane traffic.
Front Panel LED Status Array
-
Global health LED: Solid green = normal operational state, solid amber = critical hardware fault
-
Separate fault indicator LEDs for each power supply and fan tray assembly
-
Per-port dual-color link/activity LEDs for all 96 SFP+ downlinks and eight QSFP+ uplink ports
-
Remote-activatable blue UID beacon LED for fast rack asset identification during maintenance windows
4. Hardware Performance Specifications
-
Total downlink bandwidth: 96 × 10G wire-speed ports, aggregate 960 Gbps full-duplex host bandwidth
-
Total uplink bandwidth: 8 × 40G full-duplex QSFP+ ports, aggregate 320 Gbps upstream fabric bandwidth
-
Fixed oversubscription ratio: 2:1 (960 Gbps total host downlink bandwidth / 320 Gbps total fabric uplink bandwidth)
-
Non-blocking full-duplex switching fabric capacity: 2.5 Tbps
-
Wire-speed forwarding throughput: 1.4 Bpps (billion packets per second) at minimum 64-byte Ethernet frames (all ports operate full line-rate L2/L3 simultaneously)
-
Forwarding latency: Ultra-low deterministic microsecond-scale store-and-forward latency, optimized for low-jitter financial HFT and parallel HPC workloads
-
Dynamic shared packet buffer: 12 MB unified shared buffer pool to mitigate packet loss under heavy bursty virtualization and storage replication traffic
-
MAC address table capacity: 288,000 MAC entry storage
-
VLAN scaling: Up to 4096 standard IEEE 802.1Q tagged VLANs
-
Jumbo frame MTU: Configurable up to 9216 bytes for storage replication and large-scale big-data workloads
-
Spanning tree scale: 64 MSTP instances, 512 RSTP instances
-
Port-channel scaling: Maximum 64 EtherChannels, up to 16 physical ports per LACP fabric port-channel / vPC trunk
-
Zero-touch POAP provisioning: Automatic firmware synchronization, full configuration download and fleet-wide bulk upgrades triggered via a singleinstall allCLI command on the parent aggregation switch
-
Hardware lossless DCB & traffic offload features (fully controlled by NX-OS software):
-
IEEE 802.1Qbb PFC and 802.1Qaz ETS lossless Ethernet for FCoE unified storage transport
-
vPC, FEX vPC and Enhanced vPC (EvPC) redundant Layer 2 multipathing high availability
-
Four independent hardware priority queues per port with WRR/DRR traffic scheduling
-
PVLAN private VLAN isolation, UDLD link failure monitoring, broadcast/multicast storm control
-
IEEE 1588 PTP boundary clock for precise sub-microsecond time synchronization in financial co-location environments
-
Expanded SAN-specific debug counters to track FCoE pause frames and storage buffer credit loss
Physical & Environmental Parameters
-
Form Factor: 2 RU (86 mm height) standard 19-inch horizontal rack mount
-
Dimensions (H × W × D): 86 mm × 439.4 mm × 602 mm (3.39 × 17.3 × 23.7 inches)
-
Bare chassis weight (no PSUs, fans, transceivers): ~10.1 kg (22.2 lbs)
-
Fully loaded maximum weight (dual PSUs + fan tray + fully populated transceivers): ~13.5 kg (29.7 lbs)
-
Operating temperature range: 0°C ~ 40°C (32°F ~ 104°F); certified altitude up to 3,000 meters (10,000 ft)
-
Storage temperature range: -40°C ~ 70°C (-40°F ~ 158°F)
-
Operating relative humidity: 10%–85% non-condensing; storage humidity 5%–95% non-condensing
-
Compliance Certifications: FCC Class A, UL 60950-1, CE, VCCI Class A, NEBS Level 3, GR63 seismic compliance, RoHS 6 validated
AC Power Supply FRU Specifications (Sold Separately)
-
AC PSU input range: 100–240 VAC, 50/60 Hz, typical energy efficiency 89–91%
-
DC PSU input range: -40 to -72 VDC negative telecom DC power supply
-
Typical full-system power consumption: 148W with all 96 downlink ports fully populated; maximum peak power draw 413W
-
Hot-swappable OIR replacement capability; no full switch reboot required during live PSU hardware swap
-
MTBF: 291,523 hours under nominal operating conditions
5. NX-OS Software Feature Set (Managed via CLI / Cisco DCNM Graphical Platform)
Layer 2 Core Data Center Fabric Features
-
IEEE 802.1Q trunking, PVLAN private VLAN isolation (max 4096 VLANs), transparent VTP mode
-
RSTP/MSTP spanning tree suite with BPDU guard, root guard and aggressive UDLD link monitoring
-
Static & dynamic LACP EtherChannels, vPC dual-active high availability to eliminate STP blocking links and maximize uplink bandwidth utilization
-
DHCP snooping, dynamic ARP inspection, port security MAC address locking, broadcast/multicast storm control
-
DCB PFC/ETS lossless Ethernet for Fibre Channel over Ethernet unified storage convergence
-
IEEE 1588 PTP precision time synchronization optimized for low-latency high-frequency trading environments
Layer 3 Routing & Virtualization Features (Requires Optional N3K-BAS1K9 Base License)
-
Unicast routing protocols: Static routes, OSPFv2, EIGRP, RIPv2, BGP-4, 32-way ECMP multi-path load balancing
-
Multicast routing suite: IGMPv2/v3, PIM-SM, MSDP for video streaming and inter-rack storage replication workloads
-
First-hop redundancy protocols: HSRP, VRRP for gateway failover redundancy
-
Hardware-accelerated IP ACLs (4000 ingress / 1000 egress entries), route redistribution, prefix lists and route-map traffic segmentation policies
-
Maximum routing table capacity: 16,000 IPv4 prefix entries, 8,000 multicast routes
-
VXLAN bridging support for multi-tenant virtual machine mobility (software-based VXLAN implementation, no full hardware offload vs newer 3100-V models)
Security, Monitoring & DevOps Automation Tools
-
AAA authentication stack (TACACS+/RADIUS/Local user database), SSHv2 encrypted out-of-band management sessions
-
SNMP v1/v2c/v3 remote monitoring, centralized syslog logging, RMON traffic statistics collection
-
Cisco DCNM single-pane graphical fabric monitoring, automated provisioning and REST API integration for data center automation
-
NX-API REST/CLI APIs, Python scripting, Embedded Event Manager (EEM) for custom automation workflows
-
Native integration with Puppet, Chef, Ansible DevOps toolchains for infrastructure-as-code deployments
-
GOLD Generic Online Diagnostics for continuous non-disruptive hardware health validation
-
Unified bulk fleet-wide firmware upgrade via single parent switch CLI command for uniform fabric fleets
-
Real-time streaming telemetry for port latency, buffer utilization and granular error counter visibility
6. Typical Deployment Scenarios for N3K-C31128PQ-10GE Bare Chassis
-
High-density all-fiber leaf-spine data center leaf switch: 96-port 10G SFP+ fiber downlinks support dense virtualization server racks connected via Nexus 2300 unified UPQ FEX devices, merging general LAN and lossless FCoE storage traffic on a single 2RU switch tier with balanced 2:1 oversubscription for large enterprise virtualization workload fleets. Bare chassis design allows facility teams to select airflow-matched power/fan FRUs aligned with standardized cold/hot aisle rack cooling facility requirements.
-
HPC and big-data parallel compute cluster rack aggregation: Massive 96-port 10G fiber density consolidates thousands of parallel compute nodes within minimal rack space; breakout 4×10G QSFP uplink mode reduces optical hardware costs by expanding fabric port density without extra rack switches, 12 MB shared buffer absorbs massive bursty east-west cluster replication traffic.
-
Mid-market financial co-location low-latency leaf switch: Deterministic microsecond forwarding latency stabilizes trading transaction jitter, all-fiber SFP+ ports connect directly to fiber-based trading hosts, dual airflow FRU options comply with co-location facility standardized cooling rack designs, redundant hot-swappable FRUs guarantee non-stop trading fabric uptime.
-
Small-scale unified LAN/FCoE data center edge aggregation: Interoperable with Nexus 2000 blade FEX devices for mixed blade enclosure + standalone fiber server rack consolidation, supporting lossless FCoE storage transport over fiber cabling to eliminate separate dedicated SAN access switches for mid-sized storage deployments.
-
Standardized mass enterprise data center uniform fabric rollout: Bare chassis SKU separates chassis and cooling/power spare part inventory management, eliminating thermal fault incidents caused by mismatched airflow power/fan modules during on-site hardware replacement for large facility deployments.
7. Standard Package Contents of N3K-C31128PQ-10GE Bare Chassis SKU
1 × N3K-C31128PQ-10GE Nexus 31128PQ 2RU high-density all-fiber leaf bare switch chassis (no power supplies, fan tray, SFP+/QSFP+ transceivers included)
19-inch standard rack mounting rails, front cable management brackets
Blade extraction handle, locking latch, hardware installation guide, global safety & regulatory compliance documentation
Separately Mandatory Components (Not Included in Base Chassis Bundle)
-
One or two hot-swappable AC/DC power supplies (port-side intake PI or port-side exhaust PE airflow variant selected to match fan trays; dual PSUs recommended for 1+1 power redundancy)
-
One matching airflow hot-swappable fan tray (required for full system cooling; mismatched airflow fan modules trigger thermal protection shutdown)
-
SFP+ transceivers for all 96 downlink ports: 1G/10G SR/LR/ER optical transceivers, Twinax/AOC DAC fiber cables
-
40G QSFP+ transceivers, BiDi optics and 40G-to-4×10G breakout splitter cables for four fabric uplink QSFP+ ports
-
Optional feature software licenses: N3K-BAS1K9 base license required to unlock full Layer 3 routing functionality; Layer 2 switching features enabled by default without license activation
-
Auxiliary accessories: RJ45 serial console rollover cable, out-of-band management Ethernet patch cords, lockable front door, air filter panel
Short Commercial Catalog Summary for Quotation
Cisco N3K-C31128PQ-10GE is an End-of-Sale bare low-power 2RU ultra-high-density all-fiber leaf Top-of-Rack (ToR) data center switch from the second-generation Cisco Nexus 3100 data center switching platform, featuring 96 auto-sensing 1G/10G SFP+ fiber downlink ports supporting Ethernet and lossless DCB FCoE unified storage transport, eight universal 40G QSFP+ high-speed fabric uplink ports capable of native 40GE or breakout to 4×10GE fabric channels, a fixed mild 2:1 oversubscription ratio and 12 MB unified dynamic shared packet buffer optimized for high-density virtualization leaf-spine fabrics, HPC big-data clusters and mid-market financial low-latency trading workloads. Distinguished from integrated pre-populated switch variants, this-10GEbare chassis ships without factory pre-installed power supplies, fan trays or optical transceivers, allowing customers to select port-intake (PI) or port-exhaust (PE) airflow cooling FRUs and custom fiber optics aligned with standardized rack cabling and facility cooling infrastructure requirements. This stateless fixed chassis relies on separate hot-swappable power and cooling FRUs for operation; all Layer 2 switching, security, QoS and FCoE storage convergence policies are fully configured, monitored and enforced centrally on upstream Cisco Nexus parent aggregation switches running modular NX-OS. It delivers line-rate Layer 2 switching natively, optional licensed Layer 3 routing, software-supported VXLAN multi-tenant virtualization, POAP zero-touch provisioning, synchronized bulk fleet firmware upgrades and full vPC high-availability fabric topologies. Limitations include separate FRU and optical transceiver procurement mandatory for full operational readiness, larger 2RU rack footprint vs compact 1RU Nexus 31108 series leaf switches, software-only VXLAN without hardware offload, higher power consumption from full 96-port fiber density, expired official Cisco hardware support lifecycle and restricted Layer 3 routing functionality without additional base software license. Compliant with global safety, EMI, NEBS Level 3 and RoHS 6 industrial standards.
|