1.0 Executive Summary

This document provides a detailed technical analysis regarding the feasibility of utilizing a Cisco Nexus 9000 series switch to replace a Cisco MDS 9000 series dedicated Fibre Channel (FC) switch. The core conclusion is that a Nexus 9000 cannot function as a direct, like-for-like replacement for an MDS 9000 in a traditional SAN core role. However, it can serve as a powerful and effective component of a modern, converged SAN fabric by extending connectivity to the access layer using Fibre Channel over Ethernet (FCoE). This document outlines the architectural differences, technical prerequisites, and a high-level solution framework for integrating the Nexus 9000 into a SAN environment.

2.0 Extracted Technical Problem and Initial Premise

The central technical inquiry is whether a Nexus 9000 series platform, typically deployed for Ethernet data center networking, possesses the necessary capabilities to provide the full suite of services offered by a dedicated MDS 9000 multilayer director or fabric switch for a Storage Area Network (SAN).

The premise of this question is rooted in the goal of infrastructure consolidation—leveraging a single, unified fabric for both LAN and SAN traffic to reduce hardware footprint, cabling complexity, and operational overhead.

3.0 Technical Feasibility and Core Architectural Differences

The initial premise requires a critical evaluation of the fundamental technologies supported by each platform.

• MDS 9000 Series: These are purpose-built Fibre Channel switches. They operate as native Fibre Channel Forwarders (FCFs), providing the full spectrum of SAN fabric services, including Name Server, Zoning, Fibre Channel Shortest Path First (FSPF) routing, and hosting native Fibre Channel interfaces (4/8/16/32/64G FC). They form the core intelligence of a SAN.

• Nexus 9000 Series: These are primarily high-performance Ethernet switches. They do not have native Fibre Channel ports. Instead, they achieve SAN connectivity by encapsulating FC frames into Ethernet frames using the FCoE protocol. Crucially, on the Nexus 9000 platform, FCoE is primarily supported in N-Port Virtualization (NPV) mode.

The distinction between an FCF and an NPV device is the most significant factor. An NPV switch acts as a pass-through device. It aggregates multiple host-side connections (N-Ports) and presents them to the core SAN fabric through a single or small number of uplinks (NP-Ports). The Nexus 9000 in NPV mode does not perform local switching between hosts, nor does it manage the fabric’s zoning database or routing tables. All fabric logins (FLOGI) and fabric services are handled by the upstream FCF, which must be an MDS 9000 or another FCoE-capable switch operating in FCF mode (e.g., Nexus 5000/7000 series).

Therefore, a Nexus 9000 can extend a SAN fabric but cannot create or manage one on its own.

4.0 Comprehensive Solution for Integrating Nexus 9000 into a SAN

To successfully integrate a Nexus 9000 as a SAN access-layer switch, a robust solution must be built upon a lossless, converged Ethernet fabric. The following framework outlines the necessary prerequisites and high-level implementation steps.

4.1 Prerequisites

1. Hardware: A compatible Nexus 9000 model is required. This typically includes fixed-port switches like the Nexus 9300-FX/FX2/FX3 series or modular chassis with line cards supporting Unified Ports. These ports can be configured for Ethernet or FCoE operation.
2. Licensing: An appropriate SAN Enterprise or FCoE-specific license must be installed and activated on the Nexus 9000 to enable FCoE functionality.
3. Software: A stable NX-OS release with robust support for FCoE and Data Center Bridging (DCB) is mandatory.
4. Core SAN Fabric: An existing, operational SAN fabric with an FCF-capable switch (e.g., an MDS 9100/9200/9700) is required for the Nexus 9000 to connect to.

4.2 Data Center Bridging (DCB) Foundation

FCoE relies on a lossless Ethernet transport to prevent frame drops, which would be catastrophic for storage traffic. This is achieved by implementing DCB standards:

• Priority-based Flow Control (PFC) (IEEE 802.1Qbb): Provides a no-drop service for the FCoE traffic class by pausing traffic on a per-priority basis.
• Enhanced Transmission Selection (ETS) (IEEE 802.1Qaz): Guarantees a minimum bandwidth allocation for the FCoE traffic class.
• Data Center Bridging Exchange Protocol (DCBX): An extension of LLDP used to negotiate and exchange DCB capabilities and configurations between adjacent network devices.

4.3 High-Level Implementation Steps

1. Enable Features: Activate the necessary features on the Nexus 9000: feature fcoe and feature lldp.
2. Configure Lossless Network: Implement a system-level QoS policy to enable PFC for the FCoE CoS value and configure ETS for bandwidth management. Apply this policy globally or to relevant interfaces.
3. Define FCoE VLAN: Create a dedicated VLAN for carrying FCoE traffic. This VLAN must be mapped to a corresponding VSAN on the MDS core switch.
4. Create Virtual Fibre Channel (vFC) Interface: For each physical Ethernet port that will carry FCoE traffic, create a virtual Fibre Channel interface.
5. Bind vFC to Physical Interface: Bind the vFC interface to the underlying Ethernet port or Port-Channel.
6. Configure Uplinks: Configure the Ethernet interfaces connecting to the MDS core as FCoE trunks and allow the FCoE VLAN. The corresponding vFC interfaces will operate as NP-Ports.
7. Verification: Use commands such as show flogi database, show fcoe database, and show vsan to verify that end-hosts connected to the Nexus 9000 have successfully logged into the core MDS fabric.

5.0 Final Recommendation

A Nexus 9000 series switch is not a suitable replacement for an MDS 9000 at the SAN core. Instead, its strength lies in its role as an FCoE NPV access switch. The recommended architecture is a Unified Fabric model where Nexus 9000 switches provide converged LAN and SAN access at the top-of-rack, connecting servers with Converged Network Adapters (CNAs). These Nexus switches then uplink to a core MDS 9000 fabric, which continues to provide robust, intelligent, and dedicated SAN services. This design achieves the goal of consolidation at the server edge while preserving the performance and stability of a dedicated Fibre Channel core.