网络是怎样连接的 (户根勤) (Z-Library)

【One-Line Pitch】 A guided "journey" from typing a URL to seeing a rendered page, this book explains how the whole internet actually works by following one request through browser, OS, cables, switches, routers, and servers. Best for beginners and self-taught engineers who want the big picture before diving into any single protocol. 【Book Arc】 - **Opening (~0%–15%)**: Frames the whole book as a single request's journey and introduces the two halves of networking — the "transport mechanism" (packets, switches, routers) and the applications (browser, web server). Solves the "I can use the internet but don't know how it works" problem. - **Early (~15%–30%)**: Follows the browser: parsing a URL, generating an HTTP request, resolving the domain via DNS, and handing the message to the OS through the Socket library and descriptors. - **Middle (~30%–55%)**: Descends into TCP/IP inside the OS — connection setup with SYN/ACK, sequence numbers, sliding-window flow control, disconnection, then IP and Ethernet packet assembly, ARP, and how a NIC turns bits into electrical signals. - **Late (~55%–75%)**: Moves out of the machine into physical infrastructure: signal transmission over cables and hubs, switch forwarding via MAC address tables, and router forwarding via routing tables and NAT. - **Ending (~75%–100%)**: Reaches the destination — the server's NIC, IP and TCP modules, the web server program handling the request (URI mapping, CGI, access control), and the browser finally rendering the response. 【Key Takeaways】 - **The whole network is one continuous relay of packets** (Opening): every stage — browser, OS, switch, router, server — just passes a "parcel" along, which is why understanding the full path matters more than memorizing protocols. - **HTTP is only the first step** (Early): a URL must be parsed, then resolved to an IP address via DNS before any data can move, so the browser's job is far more than "sending a request." - **TCP guarantees reliable delivery on an unreliable medium** (Middle): SYN/ACK handshakes, sequence numbers, ACKs, and sliding-window buffering exist because packets can be lost, duplicated, or arrive faster than the receiver can process. - **IP and Ethernet divide labor deliberately** (Middle): IP decides *where* a packet goes; Ethernet (MAC headers) decides *how* to reach the next hop. This separation lets Ethernet be swapped for other link technologies. - **Switches learn, routers decide** (Late): switches build MAC address tables from traffic and age out stale entries; routers strip MAC headers, consult routing tables, and rewrite headers for the next hop. - **NAT is what makes home networking work** (Late): private addresses are translated to public ones at the router, which is why a single home router can hide an entire LAN behind one connection. - **The server side mirrors the client side** (Ending): receiving a request reverses the same layered process — NIC, IP, TCP, then the web server program mapping a URI to a file or CGI script. 【Reading Tips】 - Read the opening chapter fully — it establishes the "one request's journey" mental model that every later chapter builds on. - Skim the terminology columns and quiz answers on a first pass; return to them as reinforcement rather than treating them as prerequisites. - Slow down on TCP flow control and the IP/Ethernet division of labor — these are the conceptual hinges the rest of the book hangs on. - Keep the packet-diagram convention in mind (arrows drawn right-to-left to match header layouts); it will otherwise feel backwards. - Pair this with a packet-analysis tool like Wireshark to see the handshakes and headers described here in real traffic. 【Coverage Limits】 The excerpts cover the book's structure, the browser/DNS/TCP/IP/Ethernet/switch/router stages, and the server-side ending, but do not include detailed content on wireless LANs, IPv6, or security topics beyond brief mentions. Specific chapter titles beyond those shown are not covered.