Network Fundamentals and Architecture
Network Fundamentals and Architecture
This domain tests how well you can explain packet movement from application to wire and justify design choices. Expect questions on layered models, encapsulation, addressing, subnetting, and topology selection.
Core rules
- The TCP/IP model maps to OSI: application (5–7), transport (4), internet (3), link (1–2). Each layer adds or removes its own header; a switch forwards on Layer 2, a router on Layer 3.
- MAC addresses are local to a broadcast domain; IP addresses are logical and routed between subnets. ARP resolves IPv4 to MAC inside a subnet; the default gateway handles off-subnet traffic.
- CIDR notation gives prefix length. For IPv4, usable hosts = 2^(32−prefix) − 2 for traditional subnets, because network and broadcast addresses are reserved.
- Common designs: three-tier (access/distribution/core) for campus, spine-leaf for data centers, and hub-and-spoke or full mesh for WAN. Each trades cost, latency, and failure domain size.
Worked example A /24 is split into /26 subnets. Borrowed bits = 26−24 = 2, so 2^2 = 4 subnets. Host bits = 32−26 = 6, so 2^6−2 = 62 usable hosts each. Block size in the last octet is 64: .0, .64, .128, .192. The subnet 192.168.10.128/26 spans .128–.191, with usable .129–.190. This reasoning avoids memorizing a table.
Traps Do not confuse collision domains (switch ports) with broadcast domains (VLANs). Do not use all 2^host bits when the question asks for usable hosts. Latency and bandwidth are independent: a high-bandwidth link can still add delay. Finally, encapsulation order matters-payload, transport header, network header, link header.