After solving most implementation problems and making the code stable enough, I'm going to perform some practical measurments of its performance, both disk and network. I would like to investigate the influence of network congestion and high packet loss rate on the operation of the timeout-retransmit mechanism. I'm going to measure real maintenance time and cost reduction while re-installing or upgrading software on mmsnfs-ized network.
Another approach would be modifying filesystem by adding full write support with robust and reliable populating mechanism and consistent locking semantics. This should result in a multiply backed up filesystem, with efficient recovery mechanisms.