This is the first in a series of practice exams for the Cisco Data Center Foundation certification (exam code: DCFNDU). Each question includes the correct answer plus a detailed explanation — because knowing why an answer is correct matters far more than memorizing the answer itself.
These questions cover core data center concepts including three-tier vs. spine-leaf architecture, SAN design, hyperconverged infrastructure, and Cisco Unified Data Center. Whether you are studying for DCFNDU, refreshing your knowledge before a job interview, or preparing for CCNP/CCIE Data Center, this post will help.
This Practice Test Covers
Topics Covered
- Three-Tier Network Design
- Spine-and-Leaf Architecture
- Cisco Unified Data Center
- SAN and Storage Network Design
- Hyperconverged Infrastructure (HCI)
- Scaling and Redundancy
Section 1 — Three-Tier Network Design
The three-tier model (core, aggregation/distribution, access) has been the standard enterprise and data center design for decades. These questions test your understanding of which devices belong at each layer and why.
The core layer requires high-throughput, high-port-density switches that can handle aggregated traffic from the entire data center or campus. The Nexus 9500 is a modular chassis switch designed specifically for this role — it supports hundreds of 40G/100G ports and is purpose-built for core and spine roles in data centers. The Catalyst 9800 is a wireless LAN controller — while not a traditional core switch, in some campus designs it operates at the core layer for wireless infrastructure management. The Nexus 9300 is a fixed-form-factor switch more suited to the access or leaf layer due to its lower port density. The UCS Fabric Interconnect connects UCS blade servers and belongs at the access layer. A hypervisor is server software — it does not belong in any network design tier.
The classic three-tier architecture consists of: (1) Core layer — high-speed backbone, connects aggregation switches; (2) Aggregation (Distribution) layer — policy enforcement, routing between VLANs, connects core to access; (3) Access layer — connects end devices (servers, workstations, IP phones). "Spine and leaf" describes a two-tier Clos architecture used in modern data centers — it is not a three-tier design. "Physical, data link, and network" are layers of the OSI model, not network tiers.
All three switches listed — Nexus 9500, Catalyst 6800, and Catalyst 6500 — are high-capacity modular chassis platforms designed for the core layer. They provide the throughput, redundancy, and port density required to handle aggregated traffic from the entire network. The UCS Fabric Interconnect is a server connectivity device, not a network core switch. The ASA is a firewall and lives in the security layer, not the core.
Section 2 — Spine-and-Leaf Architecture
Spine-and-leaf (also called Clos architecture) is the dominant design for modern data centers. It provides predictable latency, easy horizontal scaling, and efficient east-west traffic forwarding — all critical for today's cloud and virtualization workloads.
In a spine-and-leaf architecture, every leaf switch connects to every spine switch — this is the defining characteristic. This full mesh between the two layers means any server connected to any leaf can reach any other server in exactly two hops (leaf → spine → leaf), regardless of where in the fabric they are. There are no direct connections between spine switches and no direct connections between leaf switches. This is what keeps latency predictable and uniform. The "aggregation layer" is part of the older three-tier model — it does not exist in a spine-and-leaf design.
- Scalability: Add a new spine switch → connect it to every leaf → instantly adds bandwidth across the fabric with no redesign
- Low, predictable latency: Always exactly two hops between any two endpoints in the same fabric
- East-west optimized: Server-to-server traffic (the majority in modern data centers) never needs to travel to a core router
With a single spine switch, any failure of that switch takes down the entire fabric — there is no redundancy. Two spine switches is the minimum for a redundant design. Each leaf connects to both spines, so if one spine fails, all leaf switches can still communicate through the remaining spine. In production data centers, two spines is the starting point, and four or more spines is common in large-scale deployments for added bandwidth and fault tolerance.
The Clos-collapsed core architecture — commonly called spine-and-leaf — collapses the traditional three-tier model into two layers. The spine layer replaces both the core and aggregation layers of the three-tier model, while the leaf layer replaces the access layer. This simplification reduces complexity and improves performance for modern east-west data center traffic patterns.
In a spine-and-leaf fabric, end devices connect to leaf switches, not to spine switches. When you need more ports for end devices, you add a new leaf switch and connect it upward to every spine switch. This is one of the key design advantages of spine-and-leaf — horizontal scale-out is straightforward and non-disruptive. You never connect leaf switches to each other, and you never connect spine switches directly to each other. Adding a new spine switch would add inter-leaf bandwidth, not end-device ports.
Section 3 — Cisco Unified Data Center
The Cisco Unified Data Center framework is built on three foundational pillars: (1) Unified Computing System (UCS) — converges compute, networking, storage access, and virtualization into a single cohesive system; (2) Unified Fabric — consolidates LAN and SAN traffic onto a single network fabric using technologies like FCoE, reducing cabling complexity; (3) Unified Management — provides a single management platform (Cisco UCS Manager / Cisco APIC) for the entire data center infrastructure. "Unified Access" is a campus networking concept, not a data center pillar. OTV and FabricPath are individual technologies, not framework pillars.
One of the core value propositions of Cisco Unified Data Center is the convergence of LAN (Ethernet/IP) and SAN (Fibre Channel storage) traffic onto a single unified fabric using Fibre Channel over Ethernet (FCoE). Traditionally, data centers ran two completely separate physical networks — one for data (LAN) and one for storage (SAN). This required separate cables, separate switches, separate teams, and separate budgets. Cisco Unified Fabric eliminates this separation, reducing infrastructure costs and operational complexity.
Section 4 — SAN and Storage Network Design
The two-tier SAN design uses core and edge SAN switches, similar in concept to the two-tier network model. Its key advantages are: scalability for larger environments (the core tier aggregates multiple edge fabrics), elasticity (edge switches can be added or removed without redesigning the core), and redundancy via dual-fabric (each server has two HBAs connecting to two separate fabrics — Fabric A and Fabric B — so a single switch failure never causes a storage outage). "Single point of failure" and "very expensive" are characteristics of a direct-attached or poorly designed storage setup, not of a properly implemented two-tier SAN.
A SAN (Storage Area Network) separates storage traffic from the general data network. Key benefits: Easier server maintenance — because storage is centralized on the SAN and not directly attached to individual servers, you can take a server down for maintenance without losing access to the storage data; other servers can still access shared storage. Dual-fabric redundancy — SANs are always designed with two independent fabrics (Fabric A and Fabric B). Every server connects to both fabrics, so no single switch or cable failure causes a storage outage. "Very affordable" is not accurate — SAN infrastructure (Fibre Channel switches, HBAs) is costly, which is why many organizations choose iSCSI or NFS as lower-cost alternatives.
Section 5 — Hyperconverged Infrastructure (HCI)
Hyperconverged infrastructure (HCI) by definition integrates compute, storage, and networking into a single software-defined solution running on standard x86 hardware. The Cisco + Nutanix solution combines Cisco UCS hardware (compute servers) with Nutanix software (AOS for storage, AHV or ESXi for virtualization) — making it explicitly a hardware + software solution. The hardware used is Cisco UCS rack servers, not blade servers. HCI does not use traditional SAN protocols like Fibre Channel — storage is managed entirely by the Nutanix software layer across the cluster nodes using its own distributed storage fabric.
HCI's three defining characteristics in this context: Easy expansion — add a new node to the cluster and it automatically joins the storage pool; no manual SAN reconfiguration needed. No SAN network — HCI eliminates the traditional SAN entirely; storage is distributed across the compute nodes themselves using software. Easy deployment and maintenance — HCI clusters are typically deployed in hours, not days, and managed through a single interface. "Multiple storage arrays" describes traditional SAN or NAS architecture, not HCI. "Redundant SAN switches" is again a traditional SAN concept that HCI specifically eliminates. "Fast convergence" is a routing protocol term, not an HCI characteristic.
Section 6 — Converged Infrastructure and Scaling
- FlexPod — Cisco + NetApp
- FlashStack — Cisco + Pure Storage
- Hitachi Adaptive Solutions for CI — Cisco + Hitachi
The scenario describes growth from one server to multiple servers with a centralized storage array. This is the classic use case for a SAN. A SAN allows multiple servers to share the same storage array over a dedicated, high-performance network (Fibre Channel or iSCSI). Directly attached storage cannot be shared between multiple servers. A cloud storage solution could work but introduces latency and ongoing costs not suitable for a small banking environment with on-premise requirements. A three-tier network with MDS switches is correct conceptually (Cisco MDS is a SAN switch) but is overspecified for a small environment — simple SAN is the right answer at this scale.
Key Topics Summary
| Topic | Key Fact to Remember |
|---|---|
| Three-tier architecture | Core → Aggregation → Access. Nexus 9500 / Catalyst 6800 / 6500 at core. |
| Spine-and-leaf | Every leaf connects to every spine. Two hops max. Minimum 2 spines for redundancy. |
| Scalability comparison | Spine-leaf offers ~25% greater scalability than three-tier for data center designs. |
| Cisco Unified DC pillars | Unified Computing System + Unified Fabric + Unified Management |
| LAN/SAN convergence | Cisco Unified Fabric consolidates LAN and SAN onto a single fabric using FCoE. |
| SAN dual-fabric | Every server connects to Fabric A and Fabric B — no single point of failure. |
| HCI minimum size | 3 nodes minimum. No SAN needed. Compute + storage on same nodes. |
| Nutanix layers | Cisco UCS firmware + AHV or ESXi hypervisor + Nutanix AOS storage |
How to Use This for Exam Prep
- Score yourself: 12–13 correct = Exam ready. 9–11 = Review weak areas. Below 9 = Re-study the topic sections.
- Focus on understanding why each answer is correct — the exam often rephrases questions to test the same concept differently
- Pay special attention to the "choose two" and "choose three" questions — these require complete knowledge of the topic, not just recognition of one correct answer
- The spine-and-leaf section is heavily tested — know the full mesh topology, two-hop latency, and scale-out process cold
Related Posts on Networklearner
- Data Center Foundation — Concepts and Reference Guide
- Cisco ACI Decommission Only vs Remove vs Secure Remove Explained
- How to Safely Decommission a Leaf Switch in Cisco ACI
- Why Service Graphs Matter in Cisco ACI — Complete Guide
