Final report on preservation scenarios deployed and integrated with data centres

This deliverable summarizes results of experiments that demonstrate preservation scenarios deployed and run in external data centres. In particular, it reports on large-scale video semantic analysis experiments that were introduced in Deliverable D23.1 and on medical data processing, discussed partially in Deliverable D8.3. The preservation scenarios focus on reproducibility of the experimental work. As results generally depend on the actual setting and performance of processing components, profiling of the deployed applications plays a crucial role. It is necessary to preserve not only original data, metadata, programs and the links between them, but also key characteristics provided by the profiling subsystem. The reported experimental findings prove that various features such as the allowed consumption of system resources significantly influence the quality of resulting analyses. Thus, it needs to be preserved by the deployed SCAPE system.