In this talk we examine a class of inverse problems that arise on graphs. We provide a review of recent developments, including design aspects for identifiability purposes, inference issues and applications to computer networks.