Evolvable and scalable system of content servers for a large digital preservation archives

Building a large-scale digital preservation archives system (LDPAS) for long-term access is a challenging endeavor. The three most significant system requirements contributing to that challenge are the scalability of the system, accommodation of heterogeneous digital holdings, and the evolvability of both the constituting technologies and the data formats in storage. The amount of digital-born materials requiring long-term preservation and access is overwhelming due to the fast pace of information technology and its widespread utilization in government, business corporations, academic institutions, and the general public. Compounding to the data volume challenge is the heterogeneity of the data, which range from the products of office automation and geospatial images, to multimedia artifacts. Storing and preserving digital objects must be accomplished in a way that allows access to the objects independently of the platform technology and software applications used to create these objects. As the technologies change rapidly, the LDPAS must accommodate the changes to its benefit, while also increasing in scale. A successful LDPAS must be able to adapt to the two-fold evolutionary demands. The evolution of the system software with time is the norm in any large-scale long-term system. But, even more important is the system's capability to cope with the evolution of the very software applications that generated the digital objects archived within it. In this paper, we propose an architecture of a system of Content Servers, which can scale with the data volume growth, continually sustain evolution in the face of technological change, and provide different levels of service to the different kinds of users accessing the LDPAS. We describe the concept, components and services of a Content Server in the context of digital archiving, service-oriented architecture (SOA) and the Open Archive Information System (OAIS) reference model. Several possible deployment configurations are shown to illustrate the efficient utilization of low level services and the flexibility of the system to scale up or down.