Fine-grained scalability of digital library services in the cloud

Modern digital library systems are increasingly handling massive data volumes; this content needs to be stored, indexed and made easily accessible to end users. Cloud computing promises to address some of these needs through a set of services that arguably support scalability of service provision. This paper discusses a set of experiments to assess the scalability of typical digital library services that use cloud computing facilities for core processing and storage. Horizontal scalability experiments were performed to benchmark the overall performance of the architecture with increasing load. The results of the experiments indicate that stable response times and some degree of variability are attainable due to multiple middleware servers when browsing and/or searching a collection of a fixed size. There is minimal variation in response times when varying collection sizes and equally after the caching phases. Most importantly, request sequencing proved that the quantity and age of requests have no impact on response times. The experimental results thus provide evidence to support the feasibility of building and deploying cloud-based Digital Libraries.