ADaptive DIstributed Software

Software systems and applications have become increasingly distributed, complex and dynamic, for example because of the interactions and dependencies with the physical world via various types of sensors # in the so-called Internet of Things. Another driver is the increasing interoperation between software artifacts from multiple service providers and organizations in the delivery of a single end-user service or application # service-oriented architectures in the Internet of Services. These are but two illustrations of increased distribution, complexity and dynamics. This evolution continues and the dynamics, within the software systems as well as in the environment in which systems and applications are deployed, drive the need for flexible modifications at run time. Multiple software qualities # for instance performance, security, reliability and availability # must be obtained and preserved during the life time of such dynamic distributed software systems. In fact, the required software qualities can typically not be delivered without the software system being adapted. The domain of adaptive distributed software systems is concerned with software that can react to and proactively anticipate changes. Initial research has been inspired by examples and cases that are triggered by changes in the (internal) system state # such as overloaded resources, partial system failure, etc. Meanwhile the notion of adaptive software systems has shifted and broadened: it refers to the ability of a software system to deal with immediate, short-term as well as long-term changes, and both internal and external causes of change. This evolution of the domain aligns with the evolution of distributed software systems and application domains as sketched above. The central problem in this proposal arised from the fact that distributed adaptations have become more complex, alongside the growing complexity of distributed software systems and applications and we therefore intend to improve this area by delivering new concepts, models and solutions that reach significantly beyond the state of the art. To further elaborate on this motivation, we present three observations that have become major drivers for the ADDIS research program.

Keywords:Software, ADDIS

Disciplines:Applied mathematics in specific fields