Exploiting Rateless Coding in Structured Overlays to Achieve Data Persistence

Heverson Ribeiro (1) , Emmanuelle Anceaume (1)

1 ADEPT - Algorithms for Dynamic Dependable Systems

In this paper we evaluate the performance of DataCube a P2P persistent data storage platform. This platform exploits the properties of cluster-based peer-to-peer structured overlays together with a hybrid redundancy schema (a compound of light replication and rateless erasure coding) to guarantee durable access and integrity of data despite adversarial attacks. The triptych "availability - storage overhead - bandwidth usage" is evaluated, and results show that despite massive attacks and high churn, DataCube performs remarkably well. We evaluate the performance of the rateless erasure codes implemented in DataCube. Our exploration shows how parameters selection impacts codes performance mainly in terms of decoding time, and collect strategies.

Mots clés: Peer-to-peer; Replication; Persistence; Evaluation

Domaines: Informatique et théorie des jeux [cs.GT]

Heverson Ribeiro, Emmanuelle Anceaume. Exploiting Rateless Coding in Structured Overlays to Achieve Data Persistence. 24th IEEE International Conference on Advanced Information Networking and Applications, AINA 2010, Apr 2010, Perth, Australia. pp.1165-1172. ⟨hal-00554701⟩