Authors: Sebastian Dombrowski; Tatiana Ermakova; Benjamin Fabian
Addresses: Chair of Information Systems, Humboldt-Universität zu Berlin, Spandauer Str. 1, 10178 Berlin, Germany ' University of Potsdam, August-Bebel-Str. 89, 14482 Potsdam, Germany ' Chair of Business Intelligence and Data Science, Hochschule für Telekommunikation Leipzig (HfTL), Gustav-Freytag-Str. 43-45, 04277 Leipzig, Germany
Abstract: Internet connectivity of cloud services is of exceptional importance for both their providers and consumers. This article demonstrates the outlines of a method for measuring cloud-service connectivity at the internet protocol level from a client's perspective. For this, we actively collect connectivity data via traceroute measurements from PlanetLab to several major cloud services. Furthermore, we construct graph models from the collected data, and analyse the connectivity of the services based on important graph-based measures. Then, random and targeted node removal attacks are simulated, and the corresponding vulnerability of cloud services is evaluated. Our results indicate that cloud service hosts are, on average, much better connected than average hosts. However, when interconnecting nodes are removed in a targeted manner, cloud connectivity is dramatically reduced.
Keywords: cloud computing; connectivity; availability; reliability; internet topology; graph analysis; complex networks.
International Journal of Communication Networks and Distributed Systems, 2019 Vol.23 No.1, pp.117 - 142
Available online: 07 May 2019 *Full-text access for editors Access for subscribers Purchase this article Comment on this article