Packet switching
A way to move data by chopping every message into small, separately addressed pieces — packets — and letting each one find its own way across a shared network, hopping node to node until it arrives, where the pieces are reassembled. No caller ties up a dedicated line; the network's capacity is shared, packet by packet, among everyone at once.
深層のアーカイブは当面英語で書かれています——検証済みの翻訳はロードマップに含まれています。このページではブラウザの翻訳機能がよく機能します。
✦ え、本当に？
The idea arrived twice, on two continents, from men who had never met. Paul Baran at the RAND Corporation designed it in the early 1960s so that U.S. command could survive a nuclear first strike — and AT&T's engineers told him flatly that it could not work. Donald Davies reached the same design independently in 1965 at Britain's National Physical Laboratory, and gave it the name we still use: he chose the word "packet" on a linguist's advice, because it survives translation into other languages intact.
これは何か
Packet switching is a discipline for sharing a communication network. Instead of opening a dedicated end-to-end circuit for the duration of a conversation — the way the old telephone system did — you break each message into small packets, stamp every packet with its destination address, and release them into a mesh of nodes. Each node inspects a packet's address, picks a good next hop, and forwards it, holding the packet only momentarily; different packets of the same message may take different routes and even arrive out of order, to be reassembled at the far end. The line is never reserved for anyone, so thousands of bursty conversations can interleave over the same wires, each using capacity only in the instants it actually has something to send.
なぜ重要だったのか
Circuit switching wastes a line whenever no one is talking, and it fails hard: cut the circuit and the call drops. Baran's motivation was survivability — a network with no central switchboard, in which packets route around damage, so that a command network could keep working even after much of it was destroyed. Davies's motivation was efficiency — computer traffic comes in silent-then-sudden bursts, and reserving a whole line for a terminal that types a line and then pauses is absurdly wasteful. Both arrived at the same answer: give up the dedicated circuit, share the network in tiny pieces, and push the intelligence to the edges. That answer is what made a single network able to carry many kinds of traffic cheaply, and what made it robust against the loss of any one path or node.
何を解き放ったのか
Packet switching is the switching principle beneath the ARPANET and, through it, the entire internet — a network of networks that asks of its members only that they carry packets. It is why one infrastructure can carry email, web pages, voice, and video at once, and why the loss of any single link degrades rather than halts the whole. Every technology in this corpus that assumes a universal, always-on data network — the web, streaming, cloud computing — rests on the modest, once-doubted idea that you can take a message apart, scatter the pieces across a shared mesh, and trust them to reassemble at the other end.
実用最小限の形
Store-and-forward messaging: break a message into fixed-size blocks, stamp each with a destination address, and have every node hold a block just long enough to read its address and pass it toward the target. Lose a block and only that block is resent, not the whole message.
この項目は完全な記述を待っています——地図製作者たちが作業中です。グラフ上の位置はすでに検証済みです。
解き放ったもの
最前線——その先はまだ記載されていません。
出典
- — Paul Baran, *On Distributed Communications*, RAND Corporation (1964)
- — D. W. Davies et al., 'A Digital Communication Network for Computers Giving Rapid Response at Remote Terminals,' ACM Symposium on Operating System Principles (1967)
- — Katie Hafner & Matthew Lyon, *Where Wizards Stay Up Late: The Origins of the Internet* (1996)
このページに誤りを見つけましたか？ここに記されたすべての主張は、異議に耐えるために書かれています。 訂正を提案する →