Embedding Adaptivity in Software Systems using the ECSELR framework

International audience; ECSELR is an ecologically-inspired approach to software evolution that enables environmentally driven evolution at runtime in extant software systems without relying on any offline components or management. ECSELR embeds adaptation and evolution inside the target software system enabling the system to transform itself via darwinian evolutionary mechanisms and adapt in a self contained manner. This allows the software system to benefit autonomously from the useful emergent byproducts of evolution like adaptivity and biodiversity , avoiding the problems involved in engineering and maintaining such properties. ECSELR enables software systems to address changing environments at runtime, ensuring benefits like mitigation of attacks and memory-optimization among others while avoiding time consuming and costly maintenance and downtime. ECSELR differs from existing work in that, 1) adaptation is embedded in the target system, 2) evolution and adaptation happens online(i.e. in-situ at runtime) and 3) ECSELR is able to embed adaptation inside systems that have already been started and are in the midst of execution. We demonstrate the use of ECSELR and present results on using the ECSELR framework to slim a software system.