A2.2.3 Compare and contrast networking models.
• Client-server and peer-to-peer models
• The respective benefits and drawbacks of client-server and peer-to-peer models
• Real-world applications may include web browsing, email services, online banking, file sharing, VoIP services, blockchain.
The big idea
All networked applications boil down to how many roles exist and where the state lives:
- Client-server – one or more servers hold the authoritative data and logic; clients only request or display it. (GeeksforGeeks)
- Peer-to-peer (P2P) – every participant can play both roles, exchanging data directly with other peers without a permanent central host. (CFTE)
Understanding the trade-offs between these two models lets architects pick the right foundation for everything from a home file share to a global cryptocurrency.
1 An at-a-glance comparison
| Dimension | Client-server | Peer-to-peer |
|---|---|---|
| Topology | Star / tiered hierarchy | Mesh (full or partial) |
| Control & governance | Central authority owns data and enforces policy | No single point of control; rules enforced collectively |
| Scalability | Vertical (scale-up server) or horizontal (add more servers/load-balancers) (Zealous System) | Increases organically as each new peer adds capacity—but directory / discovery may become a bottleneck |
| Reliability | High if servers are redundant; single server is SPOF | High fault-tolerance: traffic can route around failed peers, but availability depends on peers staying online (Hivenet) |
| Performance | Predictable latency; server bandwidth is limiting factor | Throughput grows with swarm size (BitTorrent) but latency may vary; NAT traversal can hurt VoIP quality (UCSB Computer Science) |
| Security | Easier to harden—central point for auth, logging, patching (Zenarmor) | Harder—no perimeter; trust established with crypto (block- chain) or reputation (P2P file sharing) |
| Management overhead | Centralised backups, monitoring, and updates | Distributed updates; no single admin view |
| Typical use cases | Web, e-mail, online banking, most SaaS | File sharing, blockchain, some VoIP/gaming, content distribution |
2 Benefits & drawbacks in depth
Client-server model
- Pros
- Centralised data management simplifies compliance, backups, and disaster recovery. (GeeksforGeeks)
- Stronger access control—all authentication pathways funnel through a known point. (Zenarmor)
- Deterministic performance—servers sit in data-centres with ample bandwidth.
- Cons
- Single point of failure (SPOF) unless replicated.
- Scales on the provider’s wallet—bandwidth and compute must grow with demand.
- Attractive attack target for DDoS or data breaches.
Peer-to-peer model
- Pros
- Cons
- Inconsistent availability—peers can vanish at any moment.
- Security is harder—must rely on encryption, consensus or reputation rather than a locked-down server.
- Governance challenges—who removes illegal content, rolls back fraud, or upgrades the protocol?
3 Real-world applications mapped to the models
| Application | Predominant model | Why it fits | Notes |
|---|---|---|---|
| Web browsing | Client-server | Browsers request pages from dedicated web servers over HTTP/S. (MDN Web Docs) | CDNs add hierarchical caching, but logic is still server-centric. |
| Client-server | SMTP relays and IMAP/POP mailboxes need central storage, spam filtering, compliance archives. | Federation (many providers) ≠ P2P; each domain still has authoritative servers. | |
| Online banking | Client-server | Regulatory audit, transaction integrity, and fraud monitoring demand central control. | Often deploys multi-tier (web-app-DB) for extra defence layers. |
| File sharing | Peer-to-peer | BitTorrent swarms slice files into pieces; download speed scales with swarm size. (Lifewire) | Legal use: Linux ISOs, game patches; risk of malware or copyright suits. |
| VoIP | Hybrid | Early Skype used super-node P2P to punch through NATs (arXiv); Zoom/Teams prefer client-server for QoS and compliance. | |
| Blockchain (e.g., Bitcoin, Ethereum) | Peer-to-peer | Consensus rules and ledgers replicated on thousands of nodes—no bank required. (CFTE) |
4 Hybrid architectures—best of both worlds
Many modern systems blend the models:
- Skype (2020-) replaced pure P2P voice paths with client-server relays for mobile reliability, while keeping P2P signalling for discovery—illustrating an adaptive mix. (Reddit)
- Content-delivery networks keep master copies on origin servers but push objects to edge peers; WebRTC’s data-channel lets browsers exchange P2P chunks inside a client-server-negotiated session.
- Bitcoin Light Clients talk to full nodes in a client-server fashion yet benefit from the underlying P2P chain.
Take-away
- Client-server centralises control, making it the go-to for transactions that demand predictable performance, tight security, and authoritative data (web shops, mail, banking).
- Peer-to-peer shines when distribution, fault-tolerance, and cost sharing outrank tight control (file sharing, public blockchains, ad-hoc VoIP).
- Real systems frequently hybridise, anchoring security and governance in servers while off-loading heavy traffic or discovery to peer meshes.
Choosing the right model—and knowing when to combine them—is a core skill for network architects aiming to meet today’s scale, reliability, and security requirements.