Cisco Nexus switches
The Cisco Nexus switches are a portfolio of modular and fixed Ethernet switches developed by Cisco Systems for data center networking, first introduced in January 2008 with the Nexus 7000 series to address the growing demands of virtualization, cloud computing, and high-bandwidth applications in enterprise environments.[1][2] These switches emphasize scalability, low latency, and resilience, supporting Ethernet speeds from 1 Gbps to 800 Gbps while integrating features like unified fabric for storage and compute convergence.[3][4] The Nexus family encompasses several series tailored to diverse data center roles, including the Nexus 3000 Series for top-of-rack, low-latency switching in high-frequency trading and cloud deployments, offering compact designs with line-rate Layer 2/3 performance and programmability via NX-OS or ACI modes.[5][6] The Nexus 5000 Series, focused on access-layer connectivity, provides high-density 10 Gigabit Ethernet and Fibre Channel over Ethernet (FCoE) support for unified storage fabrics, though many models are now end-of-sale.[7] The Nexus 7000 Series serves as a modular core/aggregation platform with up to 18.7 Tbps capacity per chassis, featuring non-blocking architecture and advanced services like VXLAN for overlay networks.[1][8] Central to the modern portfolio is the Nexus 9000 Series, a versatile lineup of fixed and modular switches that form the foundation for hybrid cloud infrastructures, supporting spine-leaf topologies, 400G/800G optics, and automation through Cisco Application Centric Infrastructure (ACI) for policy-driven management.[9][10] This series includes sub-families like the Nexus 9300 for flexible 1-800G port configurations and the Nexus 9800 for high-capacity, low-power modular systems, enabling AI/ML workloads and sustainable operations with energy-efficient silicon.[11][12] Overall, Nexus switches integrate with Cisco's broader ecosystem, including NX-OS for traditional networking and ACI for intent-based fabrics, to simplify operations and enhance security in multi-tenant environments.[3]Introduction
Background and Development
The development of Cisco Nexus switches emerged in the mid-2000s amid the rapid growth of data center virtualization, driven by increasing demands for scalable, high-performance networking to support server consolidation and virtual machine proliferation. Cisco addressed these needs through its acquisition of Nuova Systems, a startup founded in 2006 to innovate in Ethernet-based data center fabrics, initially investing as the majority stakeholder before completing full ownership in May 2008 for approximately $678 million.[13] This move aligned with Cisco's "Data Center 3.0" vision, emphasizing unified fabrics that integrated computing, networking, and storage to overcome the silos in traditional architectures.[14] The Nexus series was formally announced on January 28, 2008, with the Nexus 7000 Series as the inaugural platform, marking Cisco's entry into purpose-built data center switching.[15] Development goals centered on resolving limitations of Cisco's existing Catalyst switches, which were optimized for campus and enterprise environments but lacked the non-blocking architecture, low-latency forwarding, and high port densities required for 10 Gigabit Ethernet in dense data centers.[16] The Nexus design prioritized virtualization support, enabling seamless integration with hypervisors and dynamic workload mobility while delivering up to 1.92 terabits per second of throughput in a modular chassis.[15] Key early influences included strategic partnerships, notably with VMware, to extend networking into virtual environments. In September 2008, Cisco and VMware jointly introduced the Nexus 1000V, a distributed virtual switch that brought Cisco's security, policy enforcement, and diagnostics capabilities to VMware vSphere infrastructures, simplifying management across physical and virtual domains.[17] This collaboration underscored Nexus's role in the emerging unified computing ecosystem, later complemented by Cisco's Unified Computing System (UCS) announced in March 2009. The timeline of initial releases positioned Nexus as a pioneer in 10 Gigabit Ethernet for data centers, with the Nexus 5000 Series launched on April 9, 2008, introducing fixed-configuration switches with native Fibre Channel over Ethernet (FCoE) support to converge storage and IP traffic over a single wire.[18] Powered by the new NX-OS operating system, derived from the modular IOS but enhanced for data center reliability with features like hitless upgrades, these early models rapidly gained adoption for their innovation in low-latency, high-density connectivity.Purpose and Target Applications
Cisco Nexus switches are engineered to deliver scalable, high-performance networking fabrics within data center environments, supporting both Layer 2 and Layer 3 switching functionalities to enable efficient data movement across virtualized infrastructures.[3] These switches facilitate the construction of robust, non-blocking architectures that handle massive bandwidth demands, with port speeds ranging from 1G to 800G in the latest models, making them ideal for modern data-intensive operations.[10] Primarily targeted at data center deployments, Nexus switches excel in top-of-rack (ToR) configurations, where they connect servers directly to the network core, as well as in spine-leaf topologies that provide low-latency, east-west traffic scaling for distributed computing.[19] They are particularly suited for high-performance applications such as AI and machine learning (AI/ML) workloads, which require rapid data processing and transfer, and for hybrid cloud integrations that bridge on-premises resources with public cloud services.[4] This versatility supports seamless orchestration in environments demanding automation and policy-driven control. Key benefits include sub-microsecond latency in specialized series, such as the Nexus 3000 and 3550 lines, which achieve latencies as low as 250 nanoseconds to meet the needs of business-critical applications.[20] High port density allows for dense server connectivity in compact form factors, while energy efficiency is enhanced through silicon-based ASICs like Cisco Silicon One, which optimize power consumption with designs featuring fewer ASICs and up to 96% efficient power supplies.[21][22] In comparison to the Cisco Catalyst series, which is optimized for enterprise campus and branch networks with features like SD-Access and PoE support, Nexus switches prioritize non-blocking performance and virtualization capabilities tailored to data center demands, such as VXLAN overlays and ACI policy enforcement.[23] This distinction ensures Nexus platforms deliver superior scalability for cloud-scale fabrics without the overhead suited to general enterprise routing.[3]Core Architecture and Technologies
Hardware Design Principles
Cisco Nexus switches utilize both fixed and modular hardware configurations to meet varying demands for density, scalability, and deployment flexibility in data center networks. Fixed-configuration models, such as those in the Nexus 9300 series, integrate all ports and components into a single chassis, offering a compact, low-latency design suited for top-of-rack (ToR) applications where predictable performance and simplified management are prioritized.[24] In modular setups, like the Nexus 9500 and 9800 series, scalability is achieved through hot-swappable supervisor engines that handle control plane functions and line cards that provide high-density data plane interfaces, enabling non-disruptive expansions to support up to 288 ports or more.[25][26] This midplane-free modular architecture reduces complexity and improves upgrade paths by allowing independent replacement of components without full system downtime.[25] At the core of packet forwarding in modern Nexus switches, particularly the 9000 series, is an ASIC-based architecture powered by Cisco's custom Silicon One processors, which unify processing for routing, switching, and advanced telemetry across diverse network roles.[27] These programmable ASICs deliver high-performance, scalable throughput—such as 51.2 Tbps in the G200 variant—while supporting features like deep buffering and flow-based monitoring to handle bursty AI and cloud traffic efficiently.[27] By integrating hardware acceleration for Layer 2/3 protocols and programmability for evolving requirements, Silicon One ensures consistent, low-latency forwarding without reliance on external processors.[27] Nexus switches support an evolution of port technologies from 10 Gigabit Ethernet to 400G and 800G speeds, using QSFP28 and QSFP-DD form factors for backward compatibility and high-density interconnects.[28] These ports enable multi-rate operation, allowing seamless migration from legacy 10G/25G to modern 400G/800G links, and incorporate breakout cable support to subdivide a single port—for instance, a 400G QSFP-DD into 4x100G or 8x50G channels—maximizing flexibility in cabling and connectivity.[28] Optics compatibility extends to unified ports that handle Ethernet alongside Fibre Channel and RoCE, ensuring versatility for storage and high-performance computing environments.[4] Reliability in hardware design is enhanced by redundant, hot-swappable power supply units (PSUs) and cooling mechanisms, with configurations supporting N+1 or grid redundancy to maintain operations during failures.[29] PSUs range from 350W to 3kW capacities, often achieving over 90% efficiency to minimize heat generation and operational costs.[25] Cooling employs multiple redundant fan trays with front-to-back airflow, dynamically adjusting speeds based on temperature to ensure thermal stability, while typical efficiency metrics include less than 3.5W per 10G port under standard loads.[29][25]Software and Operating Modes
Cisco NX-OS serves as the primary operating system for Cisco Nexus switches, designed specifically for data center environments with a modular architecture that enables independent process management and restarts without requiring a full system reload.[30] Built on a Linux kernel foundation, such as Wind River Linux, NX-OS provides enhanced stability, scalability, and programmability compared to traditional networking OSes, supporting hitless upgrades that maintain traffic flow during software updates on supported platforms.[31] Additionally, it incorporates Python scripting capabilities for automation, allowing administrators to develop custom scripts for configuration, monitoring, and troubleshooting tasks directly on the device.[32] In ACI mode, available on select Nexus series like the 9000, switches operate within the Application Centric Infrastructure (ACI) framework, which shifts from traditional device-centric networking to a policy-driven model that automates provisioning and orchestration based on application requirements.[33] This mode integrates with the Application Policy Infrastructure Controller (APIC), a centralized clustered controller that enforces policies across the fabric, enabling unified management of physical and virtual resources while optimizing performance through intent-based networking.[34] ACI mode supports seamless scalability for multitenant environments, abstracting the underlying network complexity to focus on application-centric outcomes. NX-OS further enhances data center networking through support for VXLAN, which facilitates scalable overlay networks by encapsulating Layer 2 frames within UDP packets over a Layer 3 underlay, enabling virtualized workloads to span multiple physical locations without VLAN limitations.[35] Telemetry features provide model-driven monitoring, streaming real-time operational data such as interface statistics and CPU utilization to external collectors via protocols like gRPC or HTTP, allowing proactive analytics and anomaly detection in large-scale deployments.[36] Programmability is bolstered by NETCONF and YANG data models, which standardize configuration and state retrieval, supporting both native Cisco and OpenConfig models for interoperability with automation tools and orchestration platforms.[37] Security in NX-OS is tailored for data center-scale operations, incorporating role-based access control (RBAC) to define granular permissions for users and roles, restricting access to specific commands and resources based on predefined rules.[38] Access control lists (ACLs) provide fine-grained traffic filtering at Layers 3 and 4, supporting IPv4/IPv6 rules to enforce policies on interfaces and virtual routing and forwarding (VRF) instances.[39] Encryption protocols, including SSH for secure management access, TLS for API communications, and options like IPsec for tunnel protection, ensure confidentiality and integrity at data center volumes, with features such as password hashing using algorithms like SHA-256 to safeguard credentials.[40] These capabilities are compatible across Nexus hardware platforms, enabling consistent security postures from fixed to modular chassis designs.Historical Evolution
Initial Launch and Early Series
The Cisco Nexus series was first introduced on January 28, 2008, with the launch of the Nexus 7000 Series switches, marking Cisco's entry into data center-class networking platforms designed for scalability and high performance in virtualized environments. These modular switches, starting with a 10-slot chassis supporting up to 384 ports at 1/10 Gigabit Ethernet speeds, emphasized non-blocking architecture and integration with existing Cisco infrastructure.[15] Building on this foundation, the Nexus 5000 Series switches were launched on April 8, 2008, extending the portfolio to fixed-configuration designs for high-density, low-latency environments. These switches, such as the Nexus 5548P model with 48 ports, targeted 10 Gigabit Ethernet connectivity to enable efficient server-to-server, server-to-storage, and server-to-network communications, supporting the convergence of LAN and SAN traffic through features like Fibre Channel over Ethernet (FCoE). Priced starting at $36,000 for a 40-port 10 Gigabit Ethernet configuration, the Nexus 5000 was positioned to address the limitations of traditional data center architectures by delivering cut-through switching for consistent low-latency performance in 10G Ethernet deployments.[41][7][42] Early adoption of the Nexus 5000 accelerated in response to widespread data center consolidation and virtualization trends in the late 2000s, which demanded scalable, unified fabrics to reduce infrastructure complexity and costs. The switches integrated closely with Cisco's Unified Computing System (UCS), introduced in March 2009, to form a cohesive ecosystem where Nexus provided the networking backbone for UCS blade servers, minimizing cabling and adapter requirements while supporting converged I/O for storage and compute workloads. Service providers like Terremark Worldwide, StrataScale, and The Planet quickly deployed Nexus 5000 solutions in 2009 to enable advanced managed data center services, demonstrating early traction in environments requiring high-bandwidth Ethernet for storage connectivity.[41][43][44] To tackle challenges in scaling server access layers amid dense rack deployments, Cisco innovated with the Fabric Extender (FEX) concept through the Nexus 2000 Series, released on January 26, 2009, as extensions to the Nexus 5000 parent switch. These compact devices functioned as remote I/O modules or line cards, allowing a single Nexus 5000 to manage up to 576 Gigabit Ethernet server ports via reduced cabling—often using short-range twinax cables—while centralizing configuration and troubleshooting on the parent switch to simplify operations and lower top-of-rack complexity. This innovation addressed key pain points in traditional multi-switch setups, such as excessive cabling and management overhead, enabling more efficient Ethernet fabrics for storage and compute integration.[45][46][47] In the market, the Nexus 5000 positioned Cisco as a direct challenger to incumbents like Brocade in the storage networking arena, where Brocade dominated SAN directors but lacked immediate FCoE support, prompting Brocade's 2008 acquisition of Foundry Networks to bolster its Ethernet capabilities in response to Cisco's unified fabric strategy. The Nexus line received positive reception for its role in enabling data center 3.0 architectures, with rapid deployment by service providers and early hyperscale-like operators signaling strong initial impact, though detailed sales figures from 2008-2010 remain proprietary. Accompanying the hardware debut, Cisco introduced the NX-OS operating system tailored for data center reliability and virtualization.[41][44]Major Milestones and Transitions
In the 2010s, the Cisco Nexus lineup advanced significantly with the introduction of higher-speed Ethernet capabilities and software-defined networking (SDN) features. The Nexus 9000 series, launched on November 5, 2013, brought support for 10/40 Gigabit Ethernet, enabling scalable data center fabrics with enhanced performance for cloud and virtualization workloads.[48] Concurrently, the Nexus 7000 series expanded to include 40G and 100G interfaces through new line cards, addressing the growing demand for high-bandwidth aggregation in enterprise environments.[8] A pivotal transition occurred on November 6, 2013, with the launch of Cisco Application Centric Infrastructure (ACI), an SDN solution that integrated policy-based automation across Nexus switches, simplifying management and improving application deployment efficiency.[49] Entering the 2020s, the Nexus portfolio shifted toward ultra-high-speed interconnects to support AI, machine learning, and hyperscale computing. In June 2022, Cisco announced the Nexus 9800 series, featuring high-density 400G ports with scalability to 800G, powered by advanced ASICs for deep-buffered switching in AI-driven data centers.[50] This evolution included the adoption of Cisco Silicon One, a unified programmable architecture introduced to streamline routing and switching across Nexus platforms, reducing operational complexity and enabling consistent performance from access to core layers.[27] For AI-optimized applications, the Nexus 3550 series, with its FPGA-based programmable platform offering ultra-low latency and precision timestamping, saw enhancements including support for NX-OS 10.6(x) in 2025, enabling advanced features for high-frequency trading and real-time analytics use cases.[51] End-of-life announcements marked a strategic migration from legacy series to modern platforms. In April 2023, Cisco declared end-of-sale for NX-OS 7.3 on the Nexus 5500, 5600, and 6000 series, initiating a phased retirement with support extending to 2026, encouraging upgrades to the Nexus 9000 family for continued innovation and security.[52] As of 2025, future directions for Nexus switches emphasize sustainability and integrated security. Cisco's designs incorporate energy-efficient features, aligning with the company's net-zero emissions goal by 2040 through reduced power consumption in high-density 400G/800G models.[53] Zero-trust security integrations, such as those via Nexus Dashboard and Cisco Secure Access, enable policy enforcement and segmentation across hybrid environments, enhancing protection for AI workloads.[54]Product Families
Nexus 1000 and 2000 Series
The Cisco Nexus 1000V is a software-based virtual switch designed specifically for integration with VMware vSphere environments, enabling consistent network policy enforcement across physical and virtual infrastructure. Released in 2009 alongside VMware vSphere 4.0, it operates as a distributed virtual Ethernet switch, with a control plane provided by a virtual supervisor module (VSM) and data plane handled by virtual Ethernet modules (VEMs) installed on each ESXi host.[55][56] The Nexus 1000V supports up to 64 uplink ports for connectivity to physical networks and scales to thousands of virtual ports per instance, facilitating features like port profiles for simplified configuration of virtual machine networking.[56] It reached end-of-sale in August 2019, with end-of-support concluding on July 31, 2024.[57] The Nexus 2000 Series consists of fabric extender (FEX) devices that provide low-latency, passive port extension for Cisco Nexus 5000 and 7000 Series switches, allowing scalable server access without requiring active switching logic in the extender itself. Introduced in 2009, representative models include the Nexus 2148T, which offers 48 Gigabit Ethernet host ports and 4 10 Gigabit Ethernet fabric ports, and the Nexus 2248TP, supporting 48 Gigabit Ethernet Power over Ethernet ports with similar fabric connectivity.[45][58] These devices connect via high-speed fabric links to a parent switch, which manages all control plane functions, enabling a unified management domain. The series reached end-of-sale on August 24, 2022, with end-of-support scheduled for August 31, 2027.[45][59] A key feature shared by both the Nexus 1000V and 2000 Series is support for the VN-Tag protocol, a Cisco-developed extension to Ethernet framing that encapsulates virtual network tags to enable seamless integration of virtualized interfaces into the physical fabric, preserving visibility and policy application for individual virtual ports as if they were physical.[58][60] This protocol facilitates low-cost, wire-speed extension of the data center fabric while minimizing latency and operational complexity, particularly in environments blending physical and virtual workloads. As of 2025, both series are fully end-of-life, with the Nexus 1000V no longer under support and the Nexus 2000 Series entering its final years of limited maintenance for active contracts.[57][45]Nexus 3000 Series
The Cisco Nexus 3000 Series comprises fixed-configuration, low-latency switches optimized for top-of-rack deployments in data centers, delivering high-performance Layer 2 and Layer 3 switching for demanding environments.[19] Introduced on March 29, 2011, the series has undergone multiple refreshes, with recent models supporting speeds from 1G to 400G Ethernet to address evolving bandwidth needs in web-scale and service-provider networks.[5] These switches employ non-blocking architectures, ensuring wire-rate performance across all ports without oversubscription.[61] Key subseries include the 3100-V, 3200, 3400, and 3500, featuring representative models such as the Nexus 3132Q-V (48x 10G/25G SFP and 6x 40G/100G QSFP28 ports, 2.16 Tbps capacity), Nexus 3232C (32x 100G QSFP28 ports, 6.4 Tbps capacity), Nexus 3432D-S (32x 400G QSFP-DD ports, 25.6 Tbps capacity), and Nexus 3548-X (48x 10G/40G SFP+ ports, 960 Gbps capacity).[61] Some configurations, like the Nexus 31128PQ in the 3100 subseries, support up to 128x 100G ports through breakout cabling, providing scalable connectivity for dense deployments.[61] All models integrate VXLAN support for overlay network virtualization, enabling efficient multi-tenancy in cloud infrastructures.[61] The series excels in specialized features, including IEEE 1588 Precision Time Protocol (PTP) for nanosecond-level timestamping and synchronization, which is critical for high-frequency trading applications in finance.[62] It also offers compatibility with Cisco Application Centric Infrastructure (ACI) in standalone NX-OS mode, facilitating cloud-scale automation and policy enforcement.[19] Recent enhancements, such as the 3400 series models released in 2019, incorporate 400G port support to handle ultra-high-throughput workloads.[5] Actively supported with ongoing NX-OS software updates through 2025, the Nexus 3000 Series is deployed in hyperscale data centers for its sub-microsecond latencies and reliability in mission-critical operations.[63][19]Nexus 4000 Series
The Cisco Nexus 4000 Series consists of switch modules designed specifically for integration into IBM BladeCenter chassis, providing entry-level 10 Gigabit Ethernet connectivity in blade server environments. The primary model, the Nexus 4001I, is a single-height module released in 2010, featuring 14 internal auto-negotiating ports (supporting 1 Gb or 10 Gb Ethernet to connect up to seven server blades) and six external 10 Gb SFP+ uplink ports operating at wire speed. It also includes one 10/100/1000 Mb RJ-45 management port and an internal 100 Mb port for communication with the BladeCenter management module, enabling non-blocking, low-latency Layer 2 switching with a forwarding rate of 300 million packets per second. The companion Nexus 4005I model, also released in 2010, is a double-height module suited for larger BladeCenter H chassis, offering 28 internal ports for up to 14 blades and 12 external 10 Gb SFP+ uplinks to accommodate expanded connectivity needs.[64] Key features of the Nexus 4000 Series include basic Layer 2 functionalities such as VLAN support (up to 512 VLANs), Spanning Tree Protocol (IEEE 802.1D, 802.1w, 802.1s), and port aggregation via LACP, alongside Fibre Channel over Ethernet (FCoE) capabilities compliant with T11 standards, including FIP snooping and DCBX for converged networking. These switches run Cisco NX-OS software and integrate seamlessly with the Nexus 5000 Series to form unified fabrics, supporting FCoE initialization protocol snooping and compatibility with Cisco Fabric Extenders (FEX) for extended port density in small-scale data center deployments.[65] However, the series lacks native Layer 3 routing features and higher-speed interfaces like 40 Gb Ethernet, limiting its scalability to blade-specific environments rather than standalone rack deployments. Introduced as a budget-friendly option for small clusters requiring basic data center interconnectivity and storage convergence, the Nexus 4000 Series targeted environments where blade servers needed efficient 10 Gb uplinks without the complexity of full-scale modular switches. Its design emphasized simplicity and low latency for high-performance computing tasks, but the absence of advanced scalability features and support for emerging speeds rendered it obsolete for modern infrastructures. End-of-sale was announced on April 24, 2015, with end-of-support concluding on April 30, 2020.Nexus 5000 and 6000 Series
The Cisco Nexus 5000 Series switches, introduced in 2008, represent an early generation of data center Ethernet platforms designed for low-latency, line-rate 10 Gigabit Ethernet connectivity with integrated support for Fibre Channel over Ethernet (FCoE). These fixed-configuration switches targeted access-layer deployments in unified fabric environments, enabling the convergence of LAN and SAN traffic through unified ports that support both Ethernet and FCoE protocols. Key models include the Nexus 5010, a 1RU chassis with 20 fixed 10G SFP+ ports (eight of which are dual-speed 1/10G) and two expansion slots for up to 28 ports total, delivering over 520 Gbps of throughput; the Nexus 5020, a 2RU model with 40 fixed 10G ports and expansion options for up to 56 ports at 1.04 Tbps; and the Nexus 5548, a compact 1RU switch offering 32 fixed 10G ports plus one expansion slot for up to 48 ports at 960 Gbps. These models were released between 2008 and 2011, emphasizing lossless Ethernet fabrics suitable for server access and storage integration.[66][42][67][68] Building on the Nexus 5000 foundation, the Nexus 6000 Series, launched in 2011, extended mid-range capabilities to higher densities and speeds, focusing on aggregation roles with support for 10/40 Gigabit Ethernet and enhanced virtualization. These switches introduced greater scalability for growing data centers, including the Nexus 6004, a 4RU extensible fixed chassis with 48 fixed QSFP+ ports (40G) and four expansion module slots for up to 96 ports at line-rate performance exceeding 3.84 Tbps bidirectional throughput, optimized for top-of-rack or leaf-spine architectures. Another prominent model, the Nexus 5596UP (part of the evolved 5500/5600 lineup integrated with 6000 features), provides a 2RU form factor with 48 fixed unified ports and three expansion slots for up to 96 ports, achieving 1.92 Tbps throughput while maintaining FCoE compatibility. The series supports Virtual Device Contexts (VDC) for logical partitioning of the switch into multiple isolated virtual switches, enhancing multi-tenancy and resource allocation in shared environments.[69][67][70] Both the Nexus 5000 and 6000 Series share core features that advanced data center networking in the 2010s, including unified ports for seamless Ethernet and FCoE operation to consolidate I/O without dedicated SAN hardware, and Fabric Path technology for scalable, loop-free Layer 2 topologies using IS-IS routing to enable multipath forwarding without Spanning Tree Protocol limitations. These capabilities facilitated efficient traffic handling in converged infrastructures, with FCoE ensuring lossless delivery for storage protocols over Ethernet. By the mid-2020s, the series has reached end-of-life status, with end-of-sale announcements spanning 2017 to 2021, hardware end-of-support for many models on May 31, 2024, and extended security support until May 31, 2026, leaving remaining deployments primarily in legacy storage area network (SAN) fabrics where FCoE integration persists for backward compatibility.[66][71][72][52][7]Nexus 7000 Series
The Cisco Nexus 7000 Series consists of modular chassis-based switches designed for high-performance aggregation and core roles in large-scale data center environments. Introduced on January 25, 2008, the series includes models such as the 7004 (4-slot), 7009 (9-slot), 7010 (10-slot), and 7018 (18-slot) chassis, which support scalable expansion through line cards and fabric modules.[1] The Nexus 7700 Series extends this lineup with models like the 7702 (2-slot), 7706 (6-slot), 7710 (10-slot), and 7718 (18-slot), offering enhanced density for 10, 40, and 100 Gigabit Ethernet ports, with overall system throughput reaching up to 83 terabits per second (Tbps) in fully populated configurations.[22] These switches employ a midplane architecture for non-blocking fabric connectivity, enabling up to 2.8 Tbps of bandwidth per slot in advanced setups.[8] Key features include dual supervisor engines that provide 1+1 redundancy for the control plane, ensuring high availability through stateful switchover and non-stop forwarding during failures.[73] Virtual Routing and Forwarding (VRF) instances enable multi-tenancy by isolating tenant traffic at Layer 3, supporting scalable segmentation in shared infrastructures without requiring physical separation.[74] The series is ACI-ready, integrating with Cisco Application Centric Infrastructure through features like VXLAN BGP EVPN for multi-tenant fabric extension and ACI WAN Interconnect for linking non-ACI borders to ACI fabrics.[75] Recent F3-Series line cards, such as the 48-port 1/10 Gigabit Ethernet and 6-port 100 Gigabit Ethernet modules, deliver wire-rate performance and low latency, supporting high-density deployments with up to 768 native 10-Gbps ports.[76] The Nexus 7000 Series evolved from an NX-OS-only platform at launch to incorporate ACI compatibility in the 2010s via software updates, enabling hybrid deployments that blend traditional networking with policy-driven automation.[77] Partial end-of-life announcements have affected various models, including the Nexus 7700 Series, which reached end-of-sale on October 12, 2023, though core configurations remain supported until at least October 31, 2028.[78] Earlier models like the 7004 and 7010 were end-of-sale in January 2023, with support extending to February 28, 2027.[79] Today, these switches continue to serve as robust core aggregation platforms in large-scale campus and data center networks, particularly where modular scalability is prioritized, while organizations increasingly transition to the Nexus 9000 Series for cloud-native, spine-leaf architectures.[80]Nexus 9000 Series
The Cisco Nexus 9000 Series represents Cisco's premier data center switching platform, designed for high-performance, scalable networking in hybrid cloud environments. Introduced on November 5, 2013, this series combines fixed and modular switch architectures to deliver non-blocking throughput, low latency, and advanced automation capabilities suitable for enterprise, service provider, and cloud-scale deployments. Built on Cisco's proprietary silicon, the platform supports multispeed ports from 1G to 800G Ethernet, enabling seamless integration of legacy and emerging workloads while prioritizing energy efficiency and operational simplicity.[48][4] The series encompasses several subseries tailored to diverse roles in data center fabrics. The Nexus 9200 subseries provides compact, fixed-configuration switches optimized for top-of-rack access with 1G/10G/25G ports, such as the Nexus 92348GC-X model offering 48x 100M/1G Base-T ports and up to 0.696 Tbps bandwidth for cost-effective edge connectivity. The Nexus 9300 subseries focuses on high-density fixed switches for leaf and spine functions, supporting up to 400G ports; for instance, the Nexus 93360YC-FX3 delivers 7.2 Tbps throughput across 96x 100G/40G ports. Complementing these, the Nexus 9400 and 9500 subseries offer modular chassis for core and aggregation, with the 9400 providing a compact 4RU form factor for 400G densities and the 9500 enabling 800G spine deployments in scalable fabrics.[81][82][83][4] Key features include Cisco Silicon One ASICs, which power streaming telemetry for real-time visibility and analytics, enhancing congestion management and fault isolation in dynamic networks. The platform supports EVPN over VXLAN for multi-tenant fabric overlays, facilitating workload mobility and segmentation. Switches operate in dual-mode, booting either Cisco ACI for policy-driven automation or NX-OS for traditional Layer 2/3 operations. Notable models include the Nexus 9396PX, a high-port-density fixed switch that reached end-of-life status, and the Nexus 9508 modular chassis, capable of up to 153.6 Tbps system bandwidth for large-scale spines. Recent 2025 enhancements, such as the Nexus 93180YC-FX3, optimize the series for AI/ML clusters with improved buffer management and RoCEv2 support.[84][85][86][87][21][88][89] As the core of Cisco's current data center portfolio, the Nexus 9000 Series remains actively supported, with ongoing innovations for AI infrastructure and sustainability. It incorporates zero-touch provisioning through PowerOn Auto Provisioning (POAP), allowing automated image upgrades and configuration deployment to streamline initial setup and scaling. Recognized as a leader in the 2025 Gartner Magic Quadrant for Data Center Switching, the series ensures backward compatibility and future-proofing for evolving cloud demands.[90][4]Nexus 3550 and 9800 Series
The Cisco Nexus 3550 Series consists of FPGA-based switches designed for ultra-low latency applications in data centers, particularly those requiring sub-100 nanosecond port-to-port latency.[91] These platforms, including the Nexus 3550-T programmable switch and the Nexus 3550-F modular Layer 1 matrix switch, leverage field-programmable gate arrays (FPGAs) to enable custom packet processing pipelines and high-speed multiplexing. Released in 2021 with enhancements continuing into 2025, the series supports up to 48 ports at speeds including 25G and 100G Ethernet via QSFP28 and QSFP-DD interfaces, targeting workloads in artificial intelligence/machine learning (AI/ML) inference, high-frequency trading, and financial services where minimal delay is critical. Note that models such as the Nexus 3550-F and 3550-H were end-of-sale on April 10, 2023, with end-of-support on April 30, 2028.[92][93][94] Layer 1 switching latency measures as low as 3-5 ns, while Layer 3 forwarding achieves approximately 130 ns, with precision timestamping at picosecond resolution for accurate event sequencing.[95] The Nexus 3550 Series emphasizes programmability through reconfigurable FPGAs, allowing users to implement custom logic for tasks like packet filtering, load balancing, and telemetry without hardware redesigns.[92] High-density optics support scalable connectivity, with options for breakout cables to aggregate multiple lower-speed links into higher-bandwidth ports.[93] Integration with the Cisco Nexus Dashboard provides centralized management, including automation, analytics, and visibility into low-latency fabrics.[96] As an emerging solution for critical infrastructure, the series receives ongoing Cisco support, with maintenance for newer models extended beyond 2030 to accommodate long-term deployments in evolving AI and financial ecosystems, while early models like the 3550-F reach end-of-support in 2028.[97] In parallel, the Cisco Nexus 9800 Series represents a modular, high-capacity switching platform optimized for 400G Ethernet today, with architecture prepared for 800G transitions to meet growing data center demands.[50] Announced in 2022 and entering general availability in 2023, models such as the Nexus 9808 (8-slot chassis) and Nexus 9804 (4-slot chassis) deliver up to 115.2 Tbps of throughput, supporting line cards with 36-port 400G or mixed 100G/400G configurations.[98] These switches incorporate Precision Time Protocol (PTP) for sub-microsecond synchronization, enabling low-jitter performance in applications like media streaming and distributed computing where timing accuracy is paramount.[99] Multiplexing capabilities, combined with virtual output queuing (VOQ) and dynamic load balancing, minimize congestion and ensure consistent packet delivery across high-speed fabrics.[100] Key features of the Nexus 9800 Series include programmable forwarding via Silicon One ASICs, which support custom pipelines for advanced telemetry and security, alongside high-density optics for efficient 800G readiness.[50] The platform integrates seamlessly with Nexus Dashboard for orchestration, policy enforcement, and real-time monitoring, facilitating deployment in large-scale environments.[12] Positioned as an emerging standard for bandwidth-intensive workloads, the series benefits from extended support through at least 2030, including firmware updates and compatibility with future optics standards.| Feature | Nexus 3550 Series | Nexus 9800 Series |
|---|---|---|
| Architecture | FPGA-based, fixed-form factor | Modular chassis with Silicon One ASICs |
| Latency/Synchronization | Sub-100 ns (3-5 ns L1), picosecond PTP | Sub-μs PTP for low jitter |
| Port Speeds | Up to 100G (QSFP28/QSFP-DD) | 100G/400G, 800G-ready |
| Throughput | Up to 1.2 Tbps per switch | Up to 115.2 Tbps per chassis |
| Key Applications | AI/ML, high-frequency trading | High-capacity data centers, media fabrics |
| Management | Nexus Dashboard integration | Nexus Dashboard integration |
End-of-Life and Support
Discontinued Models and Timelines
Cisco has announced the end-of-life (EOL) for several Nexus switch models and series over the years, transitioning customers to newer platforms as technology evolves. These announcements typically include an end-of-sale (EOS) date, after which new orders are no longer accepted, followed by a period of continued support until the end-of-support date.[101] The Nexus 1000V virtual switch series reached EOS in 2019 for the VMware vSphere version, with end-of-support occurring on December 11, 2023. Similarly, the Nexus 2000 Series Fabric Extenders were placed at EOS on August 24, 2022, with support extending five years post-EOS in line with standard policy, until 2027. The Nexus 5000 Series, including models like the Nexus 5548P which had an individual EOS announcement in June 2016 with last order date September 26, 2015, saw the full series EOS on May 5, 2021, and end-of-support on May 31, 2026. For the Nexus 6000 Series, EOS was declared on April 30, 2017, with hardware support ending five years later in 2022; however, specific software versions like NX-OS 7.3 for Nexus 5500, 5600, and 6000 reached EOL announcement on May 5, 2020, with end-of-support on May 31, 2026.[57][45][102][7][52][103] In the Nexus 7000 Series, not all models are discontinued, but partial EOS has occurred; for instance, the Nexus 7718 as part of the 7700 switches subfamily had an EOL announcement on October 12, 2022, with last order date October 12, 2023, and end-of-support on October 31, 2028. For the Nexus 9000 Series, while most models remain active, select fixed spine switches like the Nexus 9332C reached EOS announcement in August 2023, with last order date January 30, 2024, and end-of-support on January 31, 2029. Additionally, as of 2025, select models in the Nexus 3000 Series, such as the N3K-C3432D-S (EOS March 7, 2025), and Nexus 9000 Series fixed switches like the N9K-C93180YC-FX (EOS July 31, 2024), have announced EOL. The Nexus 4000 Series, an early modular offering, was effectively discontinued prior to formalized modern EOL processes, with no active support listings post-2010s transitions to subsequent series.[79][78][87][104][105][106]| Series/Model | EOS Announcement Date | Last Order Date | End-of-Support Date |
|---|---|---|---|
| Nexus 1000V | February 2019 | August 2, 2019 | December 11, 2023 |
| Nexus 2000 Series | Various (final 2022) | August 24, 2022 | August 24, 2027 |
| Nexus 5000 Series (e.g., 5548P) | June 2016 (model-specific); May 2021 (series) | September 26, 2015 (model); May 5, 2021 (series) | May 31, 2026 |
| Nexus 6000 Series | April 2017 (hardware); May 5, 2020 (NX-OS 7.3) | April 30, 2017 (hardware) | April 30, 2022 (hardware); May 31, 2026 (software) |
| Nexus 7000 (7718/7700) | October 12, 2022 | October 12, 2023 | October 31, 2028 |
| Nexus 9000 (9332C) | August 2023 | January 30, 2024 | January 31, 2029 |
Migration and Upgrade Paths
Migrating from end-of-life (EOL) Cisco Nexus models such as the 5000 and 7000 Series to the current 9000 Series involves a combination of hardware replacement and software configuration transitions to maintain network continuity.[107] For the Nexus 5000 Series, which reached end-of-sale in 2021, and the Nexus 7000 Series, with support ending in 2026 for certain releases, direct hardware swaps to Nexus 9000 models like the 9300 or 9500 are recommended, often using virtual port channel (vPC) topologies to enable phased replacements without full network disruption.[107][108] This path preserves existing NX-OS configurations by exporting them via CLI commands likecopy running-config to a TFTP server before installing the new hardware, followed by importing and validating on the target switch.[109]
In environments leveraging Virtual Device Contexts (VDCs) on Nexus 7000 switches, migration to Nexus 9000 often incorporates software emulation through Cisco Application Centric Infrastructure (ACI) fabric modes, where VDC-based segmentation transitions to ACI's policy-driven fabrics for enhanced automation and scalability.[110] For lower-density setups, such as upgrading from Nexus 3000 Series to Nexus 9300 models, hardware swaps are straightforward, involving cabling reconfiguration and NX-OS image installation while ensuring compatibility via pre-upgrade checks.[111] These paths emphasize non-disruptive techniques, such as building parallel fabrics and gradually shifting workloads, to avoid service interruptions during the transition.[112]
Key tools supporting these migrations include the Cisco Technology Migration Program (TMP), which allows trading in eligible EOL Nexus hardware for credits toward 9000 Series purchases, facilitating cost-effective upgrades as of 2025 when many legacy models face support expiration.[113] NX-OS interoperability matrices provide verified compatibility between switch series and software releases, ensuring seamless integration—for instance, confirming NX-OS 10.x support across 7000-to-9000 transitions.[114] For ACI-integrated deployments, controller upgrades via the Cisco APIC follow a structured process: select compatible firmware versions from the support matrix, perform pre-upgrade validations, and upgrade switches in staggered groups to maintain fabric availability.[115]
Critical considerations during migrations focus on configuration preservation through automated backups and validation scripts, minimizing downtime with In-Service Software Upgrades (ISSU) on supported Nexus 9000 models, which enable image loading and supervisor switchover without traffic loss using nonstop forwarding and stateful switchover.[116] ISSU reduces upgrade windows to seconds for control plane convergence, though line cards may experience brief disruptions if not compatible.[116] Cost analyses for 2025 migrations should factor in TMP credits, potential hardware refresh cycles, and ROI from higher port speeds in 9000 Series, such as 400G/800G support, while planning for interoperability testing to avoid compatibility issues.[113][111]
A representative case study involves hyperscale data center operators transitioning from Nexus 6000 Series to Nexus 9800 Series to enable 800G Ethernet for AI/ML workloads, as outlined in Cisco's migration best practices; this shift utilized parallel fabric builds and vPC peer migrations to achieve zero-downtime cutovers while scaling bandwidth by up to 8x.[112][100]