P2P Applications

Computer Network & Network Security System — Application Layer, NEC licence examination syllabus (Nepal Engineering Council).

P2P Applications: when there's no central server at all

Every device is simultaneously a client AND a server — the exact opposite of how most of the internet works.

🔄 P2P vs Client-Server

Client-serverA dedicated server handles all requests; clients only ever talk to the server, never to each other.

P2PEvery participant ("peer") can act as both client and server — downloading from others while simultaneously uploading to others.

BitTorrent is the classic real-world example: a file is split into many small pieces, and you download different pieces from many different peers simultaneously, while also uploading pieces you already have to others — the more popular a file, the faster it downloads for everyone, since more peers are sharing pieces.

💡 One-liner: "Why does P2P scale better than client-server for very popular content?" → Load spreads across all the peers instead of hammering one central server — more downloaders actually means more available upload capacity too.

The scaling argument that justifies the architecture

P2P is not merely "no central server". It inverts the usual relationship between popularity and performance.

One file, one server, N downloaders: CLIENT-SERVER every byte comes from the server server upload capacity is fixed and shared → each new downloader makes everyone SLOWER PEER-TO-PEER each peer uploads pieces it already has total capacity grows with the number of peers → each new peer adds demand AND supply

So a client-server system is at its worst exactly when the content is most popular, while a P2P swarm gets faster as it gets busier. That single property is why BitTorrent distributes large files that would flatten a single server — and why it is used for game patches and operating-system images.

How BitTorrent makes peers cooperate

The design's real problem is not technical but social: nothing forces a peer to upload, and a swarm where everyone only downloads has no capacity at all.

🤝 The mechanisms

PiecesThe file is split into fixed-size pieces, each individually verified by hash. A peer can share pieces before holding the whole file — which is what lets capacity build immediately.

Rarest firstPeers request the piece fewest others hold, keeping every piece available and preventing the swarm dying because one piece vanished.

Tit-for-tatA peer preferentially uploads to those uploading back to it, so contributing earns faster downloads and freeloading is throttled.

TrackerHelps peers find each other. It carries no file data — losing it stops new peers joining but does not stop existing transfers.

💡 Per-piece hashing does more than integrity checking: it makes downloading from untrusted peers safe. A corrupted or malicious piece fails its hash and is discarded, so no peer has to be trusted — which is what makes a swarm of strangers workable at all.

What P2P costs

Two prices, both examinable. Finding content is harder without a central index, so systems need trackers, distributed hash tables or flooding — each with its own trade-offs. And there is no guarantee of availability: if every peer holding a file leaves, it is simply gone, where a server keeps it as long as it runs.

💡 Hence hybrids. Modern systems often keep a central component for lookup and authentication while moving the bulk data transfer to peers — the index is small and needs to be reliable, the data is large and does not need to be central.

Syllabus points

P2P architecture vs client-server
Examples (e.g. BitTorrent)

Create a free account to tick topics off, take notes as you read, watch the video lessons and get a day-by-day study plan built around your exam date.