Network Architecture and Design
Core concepts
A candidate must be able to move between the three-tier model (core, distribution, access) and its collapsed variants, explain why a two-tier spine-leaf fabric suits east-west data-center traffic, and justify redundancy choices such as device-level versus link-level versus path-level protection. Understand hierarchical addressing, summarization boundaries, and how physical topology constrains the routing design. Know the trade-offs of routed access versus switched access, and where first-hop redundancy (VRRP, HSRP) sits relative to routing convergence.
Traps in questions
Questions often present a design that satisfies availability but violates a stated constraint such as cost, management simplicity, or a maximum hop count. Watch for "best" versus "valid" wording: several options may work, but only one meets the explicit requirement. Another trap is confusing a design goal (five nines) with a mechanism (dual supervisors). Do not assume more redundancy is always correct; looped topologies without a control-plane plan create broadcast storms and slow convergence.
Worked rule of thumb
When sizing a campus, apply the 80/20 reversed-traffic heuristic loosely, then verify with measured flow data. Example: 40 access switches each offering 1 Gbps to the distribution layer, with an oversubscription target of 4:1, require roughly 10 Gbps of uplink capacity per distribution pair, so two 10 Gbps links provide both capacity and single-link failure tolerance. State the assumption explicitly in the answer rationale.
Practical guidance
Document the design intent for each layer, name the failure domain each redundancy feature protects, and check that any summarization boundary aligns with the addressing plan so that a single link failure does not black-hole a summary route.