Computer networks: principles, technologies, protocols (short summary)

A core message of the book is to treat networking as independent layers with explicit contracts.

• OSI/TCP-IP helps isolate failures to link, transport, or application level.
• Encapsulation explains header overhead and MTU constraints in real traffic.
• In system design, this maps directly to service boundaries and API contracts.

The book details how packets move through L2/L3 devices in production networks.

• Switches and routers make decisions at different layers and performance profiles.
• ARP, routing tables, and next-hop logic influence both latency and failure modes.
• Hop-by-hop visibility improves observability and incident triage.

Transport choice determines latency profile, delivery guarantees, and retry cost.

• TCP provides ordering and flow control but adds handshake and head-of-line risks.
• UDP reduces overhead and fits realtime workloads with app-level reliability.
• Windows, retransmissions, and congestion control define behavior under load.

Reliability is treated as an engineering discipline, not just protocol theory.

• Packet loss, jitter, and burst errors explain unstable p95/p99 latency.
• Timeout budgets and retry policies must be grounded in real measurements.
• RTT, loss, and retransmit telemetry are required for diagnosing degradations.

DNS and HTTP are shown as part of an end-to-end user journey, not isolated components.

• Fast backend logic cannot compensate for slow DNS/TLS phases.
• DNS/HTTP caching, keep-alive, and compression directly affect perceived performance.
• Protocol choices must account for mobile and multi-region access patterns.

Wi-Fi and mobile channels vary significantly, so network behavior is non-stationary.

• High RTT/loss variance requires adaptive timeouts and retry strategies.
• Mobile clients need graceful degradation and offline-aware design.
• This is critical for realtime and high-throughput systems.