- Atlas
- Computer Science
- Peer To Peer And Content Delivery
Peer To Peer And Content Delivery
Peer-to-peer (P2P) networking is the idea that the computers using a service can also supply it: each participant, or peer, both consumes and provides data, rather than relying on a dedicated server.
11 frameworks, from 1990 to today · Networks · Computer Science
- Client–server content delivery 1990-Present
- Content delivery network 1998-Present
- Hybrid Peer-to-Peer Indexing 1999-2001
- Anonymity and Censorship-Resistant Overlays 2000-Present
- Unstructured Overlay Flooding 2000-2010
- BitTorrent 2001-Present
- Content Adressing and Decentralized Storage 2001-Present
- Distributed hash table 2001-Present
- Peer-to-Peer Real-Time Media 2003-2015
- Peer-Assisted Content Delivery 2005-Present
- Blockchain 2008-Present
Overview
Peer-to-peer (P2P) networking is the idea that the computers using a service can also supply it: each participant, or peer, both consumes and provides data, rather than relying on a dedicated server1. Content delivery is the older, broader problem of getting the same bytes to many users quickly and cheaply. The two histories braid together because they answer the same question from opposite ends. A content delivery network solves scale by building a private, centrally managed web of servers. A peer-to-peer system solves scale by recruiting the users themselves1. What is at stake is who pays for distribution, who controls what can be published, and whether a system can keep working when its central pieces fail or are attacked.
Serving Many Users from One Place
The early World Wide Web inherited the client–server model of the networks beneath it. A browser (the client) asks a named server for a page; the server holds the authoritative copy and answers every request. This is simple and easy to reason about, but it has an obvious weakness: every user's demand lands on the same machine and the same network link1. As the web grew in the mid-1990s, popular sites buckled under traffic, and a single failure took a site offline entirely.
The first systematic answer was the content delivery network (CDN). The commercial pioneer was Akamai, founded in 1998 by Tom Leighton and Danny Lewin, building on Leighton's work on consistent hashing—a technique for spreading data across many machines so that adding or removing one moves as little data as possible2. A CDN places copies of content on servers ("edge" nodes) close to users and redirects each request to a nearby copy, typically by rewriting URLs or using DNS to hand out different addresses. The user still speaks client–server; the server is just replicated and geographically dispersed. CDNs became the default infrastructure for large-scale web distribution, and the idea of caching content near the edge remains central to how the internet delivers video and software today.
Sharing Among Equals: Napster and Central Indexing
The peer-to-peer idea arrived on the public stage with Napster in 199913. Napster did not store music files itself. It ran a central index: users' machines registered the files they held, and the index told a requesting user which peer had a given song456. The actual transfer happened directly between users' computers6. This hybrid design—central directory, distributed data—was enormously effective and enormously legally exposed, because the central index was a single, seizable point of control45. Napster was shut down by court order in 2001.
Its successors drew the lesson. Gnutella (2000) removed the central index entirely: peers formed an unstructured overlay, a network layered on top of the internet in which each peer knows a few others, and a search is propagated by flooding—each peer forwards the query to its neighbors, who forward it onward, until it reaches someone with the file or a hop limit is hit. This is robust to any single node disappearing, but it scales poorly: flooding generates traffic proportional to the number of peers, and early Gnutella networks were famously swamped by query traffic.
Swarms and Incentives: BitTorrent
BitTorrent, released by Bram Cohen in 2001, changed the economics7. A file is split into small pieces, and users downloading the same file form a swarm. Each peer downloads pieces from many others simultaneously and, crucially, uploads pieces it already has to others. This turns every downloader into a distributor, so capacity grows with popularity instead of being a bottleneck1.
BitTorrent also confronted the free-rider problem—users who download but never upload1. Its tit-for-tat rule has each peer preferentially upload to peers that upload to it, choking off those that only take. A separate mechanism, optimistic unchoking, periodically gives a new peer a chance, so that newcomers can bootstrap. The result was a system that sustained large transfers among strangers with no central server and no payment. BitTorrent became the dominant protocol for large-file distribution, and its swarming and incentive ideas are the reference point for most later P2P design.
Structured Overlays: Distributed Hash Tables
Flooding was wasteful; the next step was to make the overlay structured so that a query could be routed to the right peer in a small number of hops15. The distributed hash table (DHT) assigns each peer and each data item a key in a large identifier space, and each peer maintains links to a few others chosen so that routing converges on the peer responsible for any key. Four designs appeared almost simultaneously in 2001: CAN, Chord, Pastry, and Tapestry5. Chord, for instance, arranges peers on a circle and gives each a "finger table" of shortcuts at exponentially increasing distances, so a lookup takes a number of hops logarithmic in the number of peers.
The DHT made peer-to-peer lookup efficient and deterministic, and it became the indexing layer for later file-sharing systems (Kademlia, used by eDonkey and later BitTorrent clients, is a prominent example). It also made possible content addressing: naming data by a hash of its contents rather than by its location, so that any peer holding the matching bytes can serve them and the name itself verifies integrity89. This idea—content addressing over a DHT—underlies later decentralized storage systems8.
Anonymity, Censorship Resistance, and Real-Time Media
Because P2P traffic is visible to network operators and peers can be identified, a strand of work pursued anonymity. Freenet (2000, Ian Clarke) stored encrypted data across peers so that no one node knew what it held. The Tor network, developed from work at the U.S. Naval Research Laboratory in the mid-2000s, routes traffic through relays to hide users' locations; it is a mix of peer-operated nodes and directory authorities rather than a pure file-sharing system, but it shares the censorship-resistance goal. These systems trade performance for privacy and remain in active use.
In parallel, P2P was applied to live media1. Skype (2003) carried voice calls over a peer-to-peer overlay, using peers as relays to traverse network address translators and to distribute call setup. Early peer-to-peer streaming systems (such as CoolStreaming, 2004) delivered live video by having viewers forward streams to one another. These showed that real-time media could ride on peer cooperation, though quality-of-service and churn—peers joining and leaving—remained hard problems.
Peer-Assisted Delivery and the Hybrid Present
The most durable synthesis is peer-assisted content delivery: a CDN or publisher keeps central control and reliability but lets users' machines help serve popular content, offloading traffic to peers while falling back to servers when peers are scarce. This combines the CDN's manageability with P2P's scaling. It is the model behind many commercial video and software-distribution systems today, and it descends directly from both the CDN and BitTorrent lineages.
Blockchains and Decentralized Storage
The most recent turn applies P2P to agreement as well as to data10. Bitcoin (2008, attributed to Satoshi Nakamoto) introduced a peer-to-peer network in which nodes gossip transactions and blocks to one another and reach consensus on a shared ledger without a central authority, using proof-of-work to make tampering costly1011. This is a P2P system whose purpose is not file distribution but agreement among mutually distrusting parties10. Its gossip-and-consensus design has since been adapted by many other blockchain networks10.
Decentralized storage systems such as IPFS (InterPlanetary File System, 2014) and BitTorrent-based and blockchain-based storage projects revive content addressing: data is named by its hash and retrieved from whichever peers hold it, with the DHT as the index. These systems aim to make storage and distribution resilient to any single provider's failure or censorship12.
Where the Field Stands
Today the field is less a single technology than a set of reusable ideas. Content delivery networks remain the backbone of large-scale distribution, now often augmented by peer assistance1314. BitTorrent-style swarming and tit-for-tat incentives inform how systems motivate users to contribute. Distributed hash tables and content addressing are standard tools for decentralized lookup and storage59. Anonymity networks and blockchains carry the censorship-resistance and consensus strands110. The open questions are familiar ones: how to keep incentives fair when peers are selfish, how to guarantee quality for real-time media under churn, how to reconcile privacy with accountability, and how much of the internet's distribution can be decentralized without losing the reliability that central infrastructure provides1.
References
- Peer-to-peer (introduction, Content delivery, Resilient and scalable computer networks, Development, Incentivizing resource sharing and cooperation, Structured networks, CoopNet content distribution system, Privacy and anonymity, Unstructured networks), Wikipedia
- Akamai Technologies (introduction, History), Wikipedia
- Napster, Wikipedia
- File sharing (History), Wikipedia
- Distributed hash table (History, introduction), Wikipedia
- Peer-to-peer file sharing (History), Wikipedia
- BitTorrent (introduction, History), Wikipedia
- InterPlanetary File System, Wikipedia
- Content-addressable storage, Wikipedia
- Blockchain (introduction, History), Wikipedia
- Bitcoin protocol, Wikipedia
- Hyphanet, Wikipedia
- Content delivery network, Wikipedia
- Kollective, Wikipedia
The frameworks of Peer To Peer And Content Delivery
Each framework is a way of working the field has used: when it held, and the question it answers.
- 1990-Present
- 1998-Present
- 1999-2001
- 2000-Present
- 2000-2010
- 2001-Present
- 2001-Present
- 2001-Present
- 2003-2015
- 2005-Present
- 2008-Present

Peer To Peer And Content Delivery is a planet in Nousfarer
Its city has a tower for each of these 11 frameworks. Land, walk in, and climb one concept at a time.
Land on Peer To Peer And Content DeliveryRelated fields
More in Networks
- Congestion Control
- Network Architecture
- Network Protocols
- Networks
- Routing And Internetworking
- Software Defined Networking
- Wireless Networks
See where Peer To Peer And Content Delivery sits on the Map of Human Knowledge