The Distributed Storage Tank project extends the Storage Tank� system's application
for heterogeneous and distributed file sharing, including file sharing
in wide-area networks. The aim of this project is to allow a Storage Tank cluster to interoperate
with other Storage Tank clusters as well as with various other types of file storage systems,
such as NAS and Data Grids. Storage Tank servers in a local cluster are responsible
for caching and replication management of remote files, which allows clients to
access the data as if it were local. Distributed Storage Tank provides its clients
with a unified file access interface to networked file storage systems with SAN-level
performance and a single name space that is location transparent.
Our goal is to bring about a seamless interface for distributed and heterogeneous
file storage systems that provides performance that is comparable to local file systems.
Through cluster-managed shared caching, Distributed Storage Tank can also preserve
the bandwidth of long-haul networks and reduce the average latency of file access
because shared files need to be transported only once to each accessing site.
Grid Computing involves the sharing of a vast aggregation of geographically dispersed
data and computing resources. Efficient and scalable storage management, high performance
and easy access to geographically distributed data are often key requirements for
Grid applications. Distributed Storage Tank extends Storage Tank's benefits of high performance,
scalability and ease of management to geographically distributed data sharing.
Beyond the current focus of providing seamless and high-performance access to distributed
files based on heterogeneous systems, the Distributed Storage Tank vision includes the
incorporation of autonomic caching, replication and migration to facilitate optimized
performance in distributed file sharing, while minimizing communication and storage overhead.