General Notes
"The coolest thing that is gonna happen to your data is not going to be done by you" - Benjamin O'Steen, Open Repositories 2008
The description of the conference Open Repositories 2008
Sun Open Source lecture
- For open source, the license is the constitution of the community
- People want a vendor to get support and patches. I.e. they will not buy software just to buy it. A Open source company will be in just as great demand.
- Closed source try to turn prospects into users
- Open source try to turn users into customers
Fedora Content models
- Eddie Shin somewhat promised that the current content model system in fedora will be final in the 3.0 release.
- It is not meant to be a restriktive system, but a system for users to build their own validation schemes on.
- NSDL Ncore is performing content model validation.
- eSciDoc is performing content model validation
Fedora
- Fedora finally unveiled their roadmap
- Fedora introduces their JMS messaging system for API-M events. Gsearch currently supports this, but not much else
- Akubra Module in development.
- Deprecates LLStore
- First, only one filesystem, transactional
- Second, multible filesystems
- Honeycomb
- Entire registry in honeycomb
- Mulgara triplestore much superier to kowari.
- Relationships now reachable with the API
- 3.0 have streamlined the SQL, as much of the lag in ingest was due to the SQL statements. Reworked the tables interily.
- Atom as a serialization and deposit format
- Atom feed for fedora object
- Atom entry for fedora datastream or datastream version
- Non-fedora tools can now be used to ingest objects
- Non-fedora tools can now be used to view objects
- Atom for JMS
- Atom feed for the entire message
- Atom entry for event
- Non-fedora tools (browsers, email) can be used to view fedora events
- Muradora handles access control to fedora, beneath any GUI.
- Advanced criteria, simpler than XACML, but much faster
- Difficult to set up, but have made a Live DVD to demonstrate the system
Sun Honeycomb
- A object focused storage system, pluggable into Fedora
- Provides 16+ TB of storage, 5/2 Reed-Solomon parity, and hot-swappable disks
- Focused around Digital Objects, not files
Obsolesence systems
- APSR AONS is a system, that attempts to automate the process of discovering which formats are threatened
OAI ORE
- The ORE 0.3 spec was presented.
- beta is 04/2008
- 1.0 is the 09/2008
- The repositories cannot demand to be treated specially. The web is not about repositories, but about URLs. Things that cannot be accessed simply by URLs will be secondclass citizens on the net.
- The web have URLs for all ressources, but lack a way to designate boundaries that delineate aggreations along with and URI for these aggregations.
- Atom is one of the preferred formats for expressing these boundaries. ORE is tied to RDF, but not to any specific serialization of RDF relations, but Atom seems to be good.
Read the primer -> Resourcemap in Atom -> Abstract Datamodel -> resourcemap profile of Atom
Bit preservation
- Haber:
- PKI are the wrong tool for long term integrity
- google for "content integrity service haber", "long term integrity haber", "digital archives haber" for his solution on hash based timestamps
- Leslie Carr:
- More and more repositories are made by virtual organisations, created for some project (LHC). These will dispand quickly, and there will be no-one with the authority to take decisions for the repository.
- Have a clear plan decided for the repository before the organisation dispands.
Contacts
- Stuart haber
- Trusted Systems Lab, HP Labs
- www.hpl.hp.com/personal/Stuart_Haber
- Presented his Timestamp algorithm
- Art Pasquinellu
- Sun Microsystems, Inc.
- Education Market Strategist
- Presented the Honeycomb
- Dierk Höppner
- German National Library of Science and Technology
- Head of IT-Development
- Integrated search from several lucene instances
- Lodewijk Bogaards
- Data Archiving and Networked Services
- IT Expert
- DANS EASY based on eSciDoc
- Robin Malitz
- Humbolt University Berlin
- Interested in Lucene Indexing