README

Smache is a content-addressable storage system, originally done for a
computational biology course project.  I've since begun work on rewriting it to
distribute data (implementing quorum, etc.) as a bit of an exercise.

The intent is to maximize distribution of data (so that we can in turn,
maximumize the number of disks that spin up when you need a piece).  Unlike
other systems (map-reduce), I feel that the most reasonable fundamental
assumption that can be made for these systems moving forward is that the
network is going to be many orders of magnitude speedier than any storage
system.  Looking at the commoditization of technologies such as quad-rate IB,
this is likely to be true even in the face of advancements in solid-state
storage.