BlobSeer
========

BlobSeer is a large-scale, distributed data storage system that
addresses storage requirements in an innovating way, aiming to deal
with the explosion of data sizes in a scalable fashion.  It emerged
from the idea of leveraging versioning for concurrent manipulation of
binary large objects (BLOBs) in order to efficiently exploit
data-level parallelism and sustain a high throughput despite massively
parallel data accesses. Features of BlobSeer include:

- a versioning based access model 
- distributed data and metadata
- instant snapshots and clones
- replication and erasure code based fault tolerance support
- deduplication and compression
- multiple subprojects:
    * BSFS: enables storage for MapReduce by acting as a replacement for HDFS
    * BlobVM: enables storage and management of VMs by exposing BLOBS 
      as raw virtual machine images

BlobSeer was written by Bogdan Nicolae and forms the core of his PhD
thesis "BlobSeer: Towards efficient data storage management for
large-scale, distributed systems", defended Nov 30, 2010 at the
University of Rennes 1, France. It is currently used in numerous 
research projects of the KerData team, INRIA Rennes, France.

Parts of this code is based on the JERASURE library, Copright (C) 2007
James S. Plank and released under LGPL 2.1. Bindings for Ruby and
Fortran were contributed by Matthieu Dorier. Bindings for Python were
contributed by Benjamin Giraud.

BlobSeer is released as an open-source project under LGPL 3.0.
Please see the accompanying lgpl-3.0.txt file for licensing details.

For additional information please visit:
http://blobseer.gforge.inria.fr