LazySusan: A Flexible, Scalable Digital Repository Ingest System
MetadataShow full item record
We present LazySusan, an architecture for a digital repository ingest system designed for flexibility and performance. The elements of LazySusan are: 1) A distributed processing model - digital objects are processed and stored by multiple processes on multiple machines interacting with a central job store. 2) Scalability and flexibility - processing agents are spawned and terminated on-demand to respond to changing load conditions. 3) Storage optimzation scheduling - storage bandwidth is the main bottleneck in most large-scale digital repositories. LazySusan is designed to optimize use of the storage channel. 4) Flexible workflow - LazySusan uses a Fedora workflow datastream to track processing objects. 5) Metadata management is performed by Fedora.