DSpace on the Grid An Interim Report

1
DSpace on the Grid An Interim Report

• Richard Rodgers - MIT
• DSpace User Group Cambridge 2005

2
DSpace/SRB Integration

• Research project begun July 2004, with partners
  MIT, UCSD Libraries, SDSC and sponsored by
  National Archives and Records Administration (US)
• Improve storage scalability
• gt capacity limits on single file system 2-4TB
• gt performance I/O bottlenecks
• Facilitate/automate preservation practices
• gt primary insurance policy keep multiple
  copies, distributed geographically and
  administratively
• Realization a Grid-enabled DSpace

3
The Grid

• The Internet a network of information resources.
• The Grid a network of computational resources
  CPU, storage, bandwidth, etc
• A set of protocols and interoperation standards.
• Not just high performance computers SETI_at_Home.
• Enables utility computing.

4
The Data Grid

• Storage abstraction spinning disk, CDs, robotic
  tape systems, etc become just logical resources
  that look like ordinary files.
• Enables large-scale, distributed, local or
  remote, and hardware-independent network
  accessible storage.
• Storage as a service.

5
The SRB

• SDSC Storage Resource Broker an implementation
  of a data grid.
• Server software that can manage a collection of
  storage resources in a unified name-space.
• Open source and free for non-commercial use.
• Uses standard relational database for storage
  management the MCAT.
• Contains auditing capabilities.
• Can host multiple client systems DSpaces
  SRB native in a unified storage management
  environment.

6
Current Project Status

• Version 1.3 of DSpace (now in beta) supports SRB
  storage as an optional bitstream store.
• DSpace content can reside anywhere and can scale
  to larger collections local filesystem and SRB
  may be mixed.
• SRB-resident content can be registered into
  Dspace by supplying metadata (batch).

7
Future Work

• In version 2X, with AIPs in an assetstore,
  replication of items/collections becomes
  feasible.
• Define executable replication policies, and
  expose them in administrative UI.
• Define and integrate life-cycle operations, audit
  processes, etc on the SRB.
• Track progress or join in wiki.dspace.org