A Framework for Object Preservation in Digital Repositories

The term “digital repository” is used in disparate contexts, in both in a formal sense to describe a complex and complete preservation system and its policies, and in a more intuitive but less clearly defined sense, often implying simply a robust storage system, or a content management system. Lack of formalism in use of this term can impede development of formal requirements for building specific repository systems and, consequently, the widespread use of such systems. On the other hand, the complexity of a formal, system-wide view can impede understanding and implementation in many environments. To attempt to reconcile these concerns, we present a pragmatic definition and scope for digital repositories (systems whose primary function is the long-term preservation of digital objects). Taking a “bottom-up” approach that builds abstractions on top of reliable storage systems, we establish a minimal general vocabulary for the expression of digital objects, independent of policy, in the form of a lowlevel content model. We show that the “primitives” supplied by this vocabulary can be used to express the particular content and metadata models for several disparate repository case studies. Introduction In complement with the comprehensive top-down system perspective of the OAIS model [6] for a repository system, Rosenthal [5] discusses the characteristics of a reliable storage system, taking a “bottom-up” view of repository systems that describes storage systems as the foundational layer of the repository. Proceeding in this bottomup path from a storage system at the lowest layer, it is reasonable to define requirements for a bit-preservation system, an abstraction atop a set of robust, replicated storage systems meeting local needs. By bounding the scope of this abstracted bit layer, charging it with ensuring long-term fixity of bitstreams whose meaning is opaque to that layer, we can separate those concerns from the requirements of a system to manage the intellectual matter of digital objects (as distinct from the bits that compose the objects). The requirements for such an object preservation system can then assume that a bit preservation system underlying it is trusted to meet the requirements of fixity and security of that repository. An object preservation system can remain conceptually unaware of the characteristics of the bit preservation system’s storage layers, such as the implementation technology, replication or validation policies, or even the physical location and ownership of the storage. Definition We define a digital preservation system as a set of layered dependent systems, namely: • storage systems, • a bit preservation system, and • an object preservation system. These systems provide layers of technical infrastructure, atop which repository instances may be implemented. A repository instance comprises: • named sets of deposited digital objects, • a logical model that defines the digital objects held by the repository instance, • preservation policies and plans, • designated sets of users, and • policies and guidelines for the deposit and use of the objects by each set of users. Inside This Issue A Framework for Object Preservation in Digital Repositories .............. 1 Standards Update ..... 6