Towards Network X-ities From a Topological Point of View: Evolvability and Scalability

Network are designed not just to maximize efficiency of performance metrics, but also to enhance various X-ities: evolvability, scalability, reliability, verifiability, deployability, adaptability... As a step towards a rigorous understanding of network X-ities, this paper presents some analytic frameworks and quantitative methodologies on evolvability and scalability. The focus of this investigation in this broad and under-explored area is on the topological features of networks. We present an Evolvable Network Design (END) Tool using dynamic programming methods to design the multi-phase deployment of a network so that early phase designs are most evolvable to later phases. Then we use several quantitative measures to compare the scalability properties of star, ring, mesh, and access-aggregatorcore topologies. The purpose of this paper is to explore the important but fuzzy notions of X-ities in the relatively under-articulated area of network robustness, using simple quantitative formulations that suffice to illustrate some of the key issues at stake.