Enterprise Network Design Patterns You Actually Need to Know
Kevin — Adjacentnode
Three-tier, spine-leaf, collapsed core — what they are, when to use them, and why the textbook answer isn't always the right one in the real world.
Every enterprise network is built on one of a handful of design patterns. The names get thrown around a lot — three-tier, spine-leaf, collapsed core — but the explanations you find online are usually either too abstract or too vendor-specific to be useful.
Here's what they actually are and when you'd use each one.
Three-Tier (Hierarchical) Design
This is the classic Cisco design model: core, distribution, and access layers. Core handles high-speed switching between distribution blocks. Distribution aggregates access switches and applies policy. Access is where end devices connect.
This design made a lot of sense when networks were primarily east-west (server to client) and you had physical separation between buildings or floors. It scales well horizontally — you just add more distribution blocks. The downside is that it adds latency and complexity, and it's not great for data center workloads where you have a lot of server-to-server traffic.
Spine-Leaf
Spine-leaf is the data center design. Every leaf switch connects to every spine switch. No leaf connects directly to another leaf. Traffic between two leaf switches always goes leaf → spine → leaf, which means predictable, consistent latency regardless of where the endpoints are.
This design is built for east-west traffic — which is exactly what modern data centers have a lot of. Containers talking to containers, microservices talking to microservices. The tradeoff is that you need a lot of cables and ports, and the design doesn't extend well beyond the data center.
Collapsed Core
Collapsed core is what you use when you don't need the full three-tier model. The core and distribution layers are combined into a single layer. This is common in smaller campuses or branch offices where the traffic volume doesn't justify the complexity and cost of a full three-tier design.
It's not a compromise — it's the right design for the right scale. Running a three-tier design in a 200-person office is overengineering. Running a collapsed core in a large campus is asking for trouble.
The Real-World Answer
Most networks aren't purely one design. A large enterprise might run three-tier in the campus and spine-leaf in the data center, with the two connected through a core layer or a WAN edge. The design patterns are tools, not rules.
When someone asks you what design to use, the right answer starts with: what's the traffic pattern, what's the scale, and what's the budget? The textbook answer is always "it depends" — which is annoying but accurate.
Enjoying the content? Subscribe for weekly breakdowns.
Subscribe to Newsletter