Network Topologies
Physical topology vs logical topology: physical describes how cables and devices are physically connected; logical describes how data flows regardless of physical arrangement. Example: star physical topology where all devices connect to a switch, but the switch can create a logical bus within a VLAN.
Star: all devices connect to a central switch or hub. Failure of the central device brings down the network — single point of failure. Most common topology for LAN. Easy to add/remove devices, easy to troubleshoot.
Bus: all devices share a single cable segment. Collision domain includes all devices. Used in legacy 10BASE-2 Coaxial Ethernet. A break in the bus disrupts all communication. Rarely used today.
Ring: each device connects to two neighbors, forming a circle. Token Ring and FDDI used ring topology. Failure of one link can disrupt the ring (unless dual-ring, as in SONET). Rarely used for LAN today.
Mesh: every device connects to every other device (full mesh) or some devices have redundant connections (partial mesh). Highly resilient — multiple paths. Expensive and complex. Used in WAN environments (MPLS, SD-WAN) and data center spine-leaf. Full mesh: n(n-1)/2 connections for n nodes.
Hybrid: combines multiple topologies — a star-of-stars (hierarchical/three-tier) or a hub-and-spoke WAN with a mesh core.
Data Center and Campus Infrastructure
Three-tier hierarchical model: Core (high-speed backbone), Distribution (policy and routing), Access (end-user connectivity). Each tier has a defined role. Core: fast switching, redundancy, no policy. Distribution: inter-VLAN routing, ACLs, quality of service. Access: port security, DHCP snooping, PoE, VLANs. Provides scalability and fault isolation.
Spine-leaf architecture: data center design replacing three-tier for east-west traffic optimization. Every leaf switch connects to every spine switch. No direct leaf-to-leaf connections. Any two endpoints are always exactly two hops apart. Highly predictable latency. Scales by adding more leaf or spine switches.
Top-of-rack (ToR) vs end-of-row (EoR) switching: ToR places a switch at the top of each rack — short cable runs within the rack, fiber runs between ToR switches. EoR places switches at the end of server rows — longer per-server cable runs but fewer switches, easier management.
Power and cooling: dual power supplies, UPS (uninterruptible power supply), PDU (power distribution unit). Hot aisle/cold aisle containment directs airflow — servers face cold aisles, exhaust heat into hot aisles. PUE (Power Usage Effectiveness) = total facility power / IT equipment power — lower is better (1.0 is perfect efficiency).
Cabling Infrastructure
Structured cabling system: standardized approach to building wiring. Components: horizontal cabling (workstation to telecommunications room), backbone/vertical cabling (between floors and buildings), work area (wall outlet to device), telecommunications rooms (wiring closets), equipment rooms (data center), entrance facilities (connection to outside).
Patch panels: passive device providing termination points for horizontal cabling. Enables flexible connectivity — patch cables connect from the patch panel to the switch. 110-type punchdown blocks and keystone jacks used for termination. Using a patch panel protects switch ports from frequent cable changes.
Cable management: horizontal cable managers, vertical cable managers, cable trays, J-hooks, Velcro ties (not zip ties, which can damage cables). Proper cable management enables airflow, reduces EMI, and makes troubleshooting possible.
Physical plant documentation: cable maps, floor plans, rack diagrams, labeling conventions. Without documentation, troubleshooting is guesswork. Label both ends of every cable during installation.