Skip to Main Content
In this paper we present two alternative ways of implementing a meta-data management subsystem for the N3FS distributed file system using Chord. In order to obtain high availability in presence of node failures and network losses, we develop solutions based either on replication or on redundancy. In order to assess the relative merits of these alternatives in terms of performance and availability, we performed an extensive simulation analysis. Our preliminary results indicate that, while there is not a clear winner in terms of absolute performance, the redundancy-based solution is characterized by an availability higher than the replication-based counterpart, especially in scenarios when node failures or network losses increase.