Towards highly adaptive data-intensive systems: A research agenda

Data-intensive software systems work in different contexts for different users with the aim of supporting heterogeneous tasks in heterogeneous environments. Most of the operations carried out by data-intensive systems are interactions with data. Managing these complex systems means focusing the attention to the huge amount of data that have to be managed despite limited capacity devices where data are accessed. This rises the need of introducing adaptivity in accessing data as the key element for data-intensive systems to become reality. Currently, these systems are not supported during their lifecycle by a complete process starting from design to implementation and execution while taking into account the variability of accessing data. In this paper, we introduce the notion of data-intensive self-adaptive (DISA) systems as data-intensive systems able to perform context-dependent data accesses. We define a classification framework for adaptation and we identify the key challenges for managing the complete lifecycle of DISA systems. For each problem we envisage a possible solution and we present the technological support for an integrated implementation.