Our notion of IT infrastructure is changing dramatically. We used to be happy with just a power plug for our desktops - today, many of us can hardly imagine a day without Internet access. Data centres provide storage and other dedicated services to large organizations - today, anybody can access such services in the cloud. Grid technologies offer a complex environment for compute-intense applications . . . however, we still lack a simple, yet trustworthy environment for curating, semantically modelling, and sharing file-based information. This thesis develops the foundational concepts for such an environment, and it builds on a class of systems stemming from scholarly communications called "digital repositories". Rather than building a new system, we analyse the interoperability channels between existing repositories, added-value services, and other agents. The Open Repository Environments emerging from these gateways are evolutionary in that repository components can change over time; anarchic to the point where they invite decentralised agents to participate; and a bridge between existing infrastructures. The infrastructures that were of particular influence to this thesis are e-Infrastructure for virtualising heterogeneous resources, as well as repository technologies for managing reference networks of file-based information, namely "digital objects". Like any software architecture, the design of Open Repository Environments is a creative process that may be inspired by existing models and experiences, and this thesis creates a framework for such models. Its repository reference architecture derives the minimum consensus on what repositories constitute from an extensive analysis of existing experiences, and specifies two interfaces through which interoperability channels can be established: the Open Storage Interface and the Federation Interface. Subsequently, we present results from a detailed analysis of the functionalities and design criteria for an Open Storage Interface. On the other hand, Federation is a continuum of attributes and techniques rather than a single interface. In our analysis of Federation, we describe this continuum, identify existing federation patterns, and develop a novel Notification-based federation mechanism - probably the first such attempt since the creation of the OAI Protocol for Metadata Harvesting in 2001. This thesis builds on a number of linked activities that were mostly conducted between 2001 and 2009, and in participation of six national and international research projects in the field. These activities combine analysis and discussion, notably in a series of workshops we organised in Europe and the US to bridge the e-Infrastructure, preservation and repository communities, as well as the architectural design and prototyping of the e-Infrastructure initiatives TextGrid and Dariah based on the findings presented in this thesis.
Read more