SPIRE cluster summary

From DigiRepWiki

SPIRE

The SPIRE project is looking into the feasibility of using peer-2-peer systems in UK academic settings. We are focusing on the LionShare open source P2P system which has been created by Penn State University. The LionShare client application differs from normal P2P clients in that it authenticates with your institutions central security system and it can also use aspects of Shibboleth to create access controls for selected files.

The main areas of interest we have as a project are:

  • Infrastructure: How does an informal technology such as P2P fit into and integrated infrastructure?
  • Shibboleth: When will large federations be in place? How could a federation use Creative Commons?
  • Context / Annotation: How can context be given to digital objects in repository style systems.
  • Workflow: Can the informal P2P philosophy blend into current workflow and institutional requirements?
  • Uptake: Would people use an informal P2P system if it was formally rolled out by the institution?

A more comprehensive list of the areas we are looking into or that have arisen are as follows:

Is it possible to transfer a dynamic and freeform technology from the wider web and place it in an academic environment in a meaningful way?

How would students and academics use P2P as part of their studies / research?

Do the formal requirements of the institution negate the initial flexibility of the P2P format? Is the reason people use P2P partiality due to anonymity?

Will the Virtual Home concept work for users who are not members of an institution?

Context:

  • In an academic environment, often the context that comes with an object (picture, document etc) is as, if not more important than the actual object its self. The SPIRE project feel that some sort of annotation / contextualisation system should be attached to each object which could potentially start a useful commentary.

IP

  • It would appear that the P2P network could only extend as far as any IP agreement that was in place between institutions. We envisage that this agreement could be attached to the Shibboleth federation that the P2P system was linked with.
  • Because of the informal nature of P2P we envisage that a creative commons licence would need to be applied to all objects within the P2P system by default.

Shibboleth

  • To account for IP (see above) and to allow the use of access controls on shared files LionShare needs to be part of a Shibboleth federation. As such the SPIRE project is interested in the general uptake of Shibboleth.

Uptake

  • A P2P network needs a critical mass of users and materials before is becomes popular. If P2P is appropriate how do we generate initial interest?
  • Because it is technically challenging to tie any technology to centralised security systems any authenticated P2P system would have to be rolled out by a central IT department. This means that what is normally a viral model of uptake for P2P would have to change to a centralised rollout.
  • LionShare also has a federated repository search. Would a client app rolled out as part of an institutional desktop be more popular than a website for a single point of contact search?