Network Topologies and Design Patterns
Key Takeaways
- Star topologies connect endpoints to a central device and are common in Ethernet access networks.
- Mesh improves resiliency by adding alternate paths, but full mesh becomes expensive and complex.
- Spine-leaf designs provide predictable east-west data center paths with equal-cost links.
- Three-tier campus design separates access, distribution, and core functions; collapsed core combines distribution and core.
- Hub-and-spoke and point-to-point describe WAN or site connectivity patterns that affect path selection and failure impact.
Topology questions ask how devices are connected and what happens during growth or failure. The best answer usually follows the traffic pattern and resilience requirement.
| Topology | What it means | Strength | Weakness |
|---|---|---|---|
| Star | Endpoints connect to a central switch or hub | Simple, common, easy to manage | Central device is critical |
| Full mesh | Every node connects to every other node | Maximum path redundancy | Expensive and complex |
| Partial mesh | Some redundant interconnections | Balanced resilience and cost | Requires routing design |
| Point-to-point | One direct link between two nodes | Simple dedicated path | Link failure isolates the path |
| Hub-and-spoke | Branches connect through a central hub | Centralized control | Hub can bottleneck or fail |
| Spine-leaf | Every leaf connects to every spine | Predictable data center paths | Requires consistent design and routing |
Campus Designs
Traditional campus architecture often uses three tiers. Small or medium networks may collapse the core and distribution layers.
| Layer | Role | Common devices |
|---|---|---|
| Access | Connects endpoints | Access switches, APs, phones, cameras |
| Distribution | Aggregates access and applies policy | Layer 3 switches, ACLs, routing boundaries |
| Core | Fast transport between distribution blocks | High-speed redundant switches |
| Design | Best fit | Exam clue |
|---|---|---|
| Three-tier | Large campus | Separate access, distribution, and core |
| Collapsed core | Smaller campus | Distribution and core combined |
| Spine-leaf | Data center | East-west traffic, leaf uplinks to every spine |
Spine-Leaf Thinking
In a spine-leaf fabric, servers connect to leaf switches, and leaf switches connect to spine switches. Leaves do not usually connect directly to other leaves, and spines do not usually connect directly to other spines. Equal-cost multipath can distribute traffic across multiple spine links.
| Traffic pattern | Design concern |
|---|---|
| North-south | Client to server, internet to data center, branch to app |
| East-west | Server to server, microservices, storage replication |
| Predictable latency | Spine-leaf can keep hop count consistent |
| Fast growth | Add leaf switches for ports, add spine capacity for bandwidth |
WAN Patterns
Hub-and-spoke is common when branches use a headquarters or cloud security stack as the central path. Point-to-point is a direct site-to-site link. Mesh WAN gives sites direct paths but increases complexity.
| Scenario | Likely pattern |
|---|---|
| Branches send all traffic through headquarters | Hub-and-spoke |
| Two data centers have a direct private circuit | Point-to-point |
| Every branch can connect directly to every other branch | Mesh |
| Small office switches connect endpoints to one access switch | Star |
PBQ-Style Thinking
Scenario: A data center application has heavy server-to-server traffic and unpredictable bottlenecks through aggregation switches. A spine-leaf design is likely because it gives leaf switches consistent uplinks to all spines and supports equal-cost paths.
Scenario: A small campus has two access switch closets and one pair of redundant Layer 3 switches doing policy and fast transport. This is closer to collapsed core than a full three-tier design because distribution and core roles are combined.
Failure Impact Drill
| Failure | In a simple design, what breaks? | Mitigation idea |
|---|---|---|
| Central switch in star fails | Attached endpoints lose connectivity | Redundant access design where justified |
| Hub site in hub-and-spoke fails | Spokes may lose shared paths | Redundant hubs or direct backup paths |
| One point-to-point link fails | Two endpoints lose that path | Secondary circuit or alternate routing |
| One spine fails | Leaves still use remaining spines | Redundant spine switches and ECMP |
A data center design connects every leaf switch to every spine switch to support predictable east-west traffic. Which topology is this?
A small campus combines distribution and core functions into the same redundant switch pair. Which design is described?
Which statements about hub-and-spoke WAN design are accurate? Choose two.
Select all that apply