Biologically-inspired design: getting it wrong and getting it right

Abstract

Large, complex computing systems have many similarities to biological systems, at least at a high level. They consist of a very large number of components, the interactions between which are complex and dynamic, and the overall behavior of the system is not always predictable even if the components are well understood. These similarities have led the computing community to look to biology for design inspiration. But computing systems are not biological systems. Care must be taken when applying biological designs to computing systems, and we need to avoid applying them when they are not appropriate. We review three areas in which we have used biology as an inspiration to understand and construct computing systems. The first is the epidemiology of computer viruses, in which biological models are used to predict the speed and scope of global virus spread. The second is global defenses against computer viruses, in which the mammalian immune system is the starting point for design. The third is self-assembling autonomic systems, in which the components of a system connect locally, without global control, to provide a desired global function. In each area, we look at an approach that seems very biologically motivated, but that turns out to yield poor results. Then, we look at an approach that works well, and contrast it with the prior misstep. Perhaps unsurprisingly, attempting to reason by analogy is fraught with dangers. Rather, it is critical to have a detailed, rigorous understanding of the system being constructed and the technologies being used, and to understand the differences between the biological system and the computing system, as well as their similarities.