Distributed Encrypted P2P Storage System with Deduplication

Overview

Bachelor’s thesis in Computer Science at FOM University (final grade 1.0). The goal was to design, implement, and analyse an experimental prototype to evaluate the feasibility of a distributed, encrypted peer-to-peer storage system with deduplication and its potential for secure, decentralised data storage.

Key Technical Findings

Convergent encryption

Convergent encryption enables content-based deduplication without exposing plaintext data, constructively resolving the inherent tension between security and storage efficiency.

Resilience

The prototype demonstrated strong resilience against partial node failures within the cluster. Individual node outages did not critically impair data availability.

Deduplication paradox

Despite reducing redundant data transfer, deduplication increases the overall storage consumption of the cluster, a trade-off explicitly measured and discussed in the thesis.

Metadata overhead

In the prototype, metadata accounts for approximately 40% of consumed storage space, significantly constraining the payload-to-metadata ratio. This effect grows with dataset size and points to concrete areas for future research and development.

Assessment

The research confirmed that a functional, secure P2P storage system with deduplication is feasible in Go. Convergent encryption, fault tolerance, and deduplication were successfully combined, and the thesis was awarded a final grade of 1.0.