As already mentioned, we are convinced that future distributed architectures will see increasingly more non-hierarchical interactions taking place between constitutive units. From our point of view, reversing that trend is neither possible nor desirable. Not only does it open fascinating perspectives in terms of better resource management [17, 18], but also has very positive implications for network robustness and survivability, as previously discussed in the case of the hypergrid.
So far, we have shown that a decentralised topology (i.e. featuring homogeneous node degree) is less prone to cascade failure and therefore better equipped to maintain quality of service when damaged than a hierarchical architecture. We have also presented strong evidence that, in theory, an adaptive perimeter defence protocol is considerably more capable than the traditional firewall strategy when it comes to resisting a dynamic threat like a propagating malicious entity. But the real challenge and the main purpose of the work described in this chapter is to demonstrate that these two ways of improving network survivability are actually complementary, and are most successful when combined into an integrated approach to information security.
As traffic and/or processing is increasingly more evenly distributed, large systems become less sensitive to a directed attack, which can be regarded as the trivial counterpart to the results of Albert et al [5] on scale-free networks. So if future designs do indeed belong to this category, this particular threat is likely to progressively fade...
TABLE OF CONTENTS 
