[Part 6] Six modern software architecture styles: P2P

This blog post series reviews six common architectural styles used in distributed systems and talks about how to choose the best one for your use case.

This blog post focuses on the Peer-to-Peer computing architectural style, however we also recommend checking the other posts in the series:

(1) Monolithic
(2) Microservices
(3) Event-Driven
(4) Serverless
(5) Edge Computing
(6) Peer-to-Peer

Peer-to-Peer Computing Architectural Style

One-liner Recap:

Liberté, Égalité, Fraternité for all the peers!

Or, technically speaking, decentralized communication and resource sharing directly between devices.

Background:

It can be argued that the basic concepts of Peer-to-Peer (P2P) computing had already been envisioned starting from the earliest discussions on software systems and networking. In fact, it’s mentioned in the RFC 1 “Host Software” by Steve Crocker and it’s echo’ed in Tim Berners-Lee's vision for the World Wide Web.

P2P is a concept that sparked conversations outside of software engineering (with heavy social and political implications) and it’s still today one of the core topics in the network neutrality controversy. It’s not difficult to see why since it’s based on the idea of decentralization, promoting a more “egalitarian” model where each node (peer) is an equal participant in the network.

In contrast to the classic centralized client/server architecture, in P2P, nodes play a dual role, acting as both a client and a server. This means that peers can directly interact with each other, sharing resources such as processing power, disk storage, and network bandwidth, without the need for a central server.

Although P2P had previously been used in many application domains, it’s the rise of the file-sharing app Napster in 1999 that brought it to the attention of the wider public. Both for the innovative idea of sharing music files for free with strangers and the legal controversy that followed.

P2P Architecture is still very much relevant today: it’s a complementary and critical technology for building scalable distributed systems, especially when it comes to edge computing applications. By integrating P2P into Edge computing architectures, you leverage its established communication model to facilitate direct communication and data sharing, thereby enhancing interoperability and enabling real-time, decentralized decision-making.

Weaknesses:

Embracing P2P architectures inherently presents a complexity challenge due to the decentralized nature of this type of network. It’s not just a question of scale due to the number of nodes and the challenge of coordinating and managing communication among them. It’s also complicated by the dynamic nature of P2P architectures, with peers joining and leaving frequently.

The management of network topology requires robust algorithms for discovering, connecting, and maintaining peer relationships (and robust tooling for documenting the system architecture).

This type of architecture also presents security challenges: without a central authority to manage security, P2P networks can be vulnerable to attacks (e.g. each peer is responsible for implementing their own layer of network security, and peers themselves might share malicious files or launch attacks against other peers).

The decentralized nature of P2P also creates unique observability challenges. Unlike traditional architectures where you can monitor centralized servers, in P2P systems, understanding what's happening requires visibility across potentially thousands of autonomous peers. When users report issues, determining whether the problem stems from their local peer, their connections to other peers, or the broader network topology becomes extremely difficult.

For applications that include traditional client-server components alongside P2P functionality (such as authentication services, discovery servers, or coordination layers) full stack session recordings can capture the centralized portions of user interactions, helping teams debug the boundaries where users transition between client-server and peer-to-peer modes. However, capturing complete visibility across fully decentralized P2P networks remains an open challenge.

Not to mention that P2P systems can face more legal and ethical issues than centralized or client-server systems due to the lack of a central authority or responsibility.

Strengths:

On the positive side, P2P architectures boast enhanced resilience, scalability, fault tolerance, and reduced dependency on a central point of failure. Also, since each peer can act as a server, the system has more capabilities than a client-server architecture. Each server peer adds its capabilities to the overall system.

Real-world use cases:

The P2P Architectural Style is used for many types of applications, with content distribution being the most popular (e.g file-sharing à la Napster). This includes software publication and distribution, content delivery networks, streaming media and peercasting for multicasting streams.

An interesting project in this space was SETI@home. Until March 2020, it used crowdsourced computers to analyze radio signals with the aim of searching for signs of extraterrestrial intelligence.

Besides content sharing, other applications of the P2P architecture typically see VoIP / communication platforms (e.g. Skype), blockchain technology (e.g. Bitcoin), distributed Machine Learning, and distributed social networks (although this is not to be confused with the “fediverse” conversation).