Diversity: A Heuristic to Improve Robustness of Self-Adaptive Cloud Architectures

In complex biological systems, the hypothesis that bio-diversity contributes to stability or robustness is an active debate. The FP7 DIVERSIFY project tests whether this hypothesis holds for software systems, and explores the use of diversity as a heuristic to increase robustness in self-adaptive architectures. Inspired by Ecology, we present here a technique to evaluate diversity of software architectures and we report preliminary investigations of its correlation with robustness. Given existing cloud-based architectures, we artificially inject predefined levels of diversity and measure the resulting robustness. In four out of our five industrial case studies, a higher diversity appeared correlated with a higher robustness.