network as to support a given traffic demand matrix and satisfy some additional
constraints (integrality, unsplittable flows, resilience etc.).
Nevertheless, in many settings we have to support a set X of (non-simultaneous) traffic matrices:
this set can be given either explicitely (e.g. day/night traffic matrix) or
implicitely (e.g. by a set of constraints defining a traffic polytope).
A sensible question in tis scenario is the following: is there any
matrix M in the set X dominating all the others, i.e. any capacity
reservation supporting M also supports all the others?
In the talk we show a good characterization for domination, discuss some
extensions and show relationships with the so-called VPN/Hose problem.