Clouds-of-clouds for dependability and security: Geo-replication meets the cloud

Abstract

The complexity of large cloud offerings makes it extremely hard to guarantee their dependability and security. This paper extracts lessons from some years of research on the notion of using several clouds -instead of a single one- with the objective of achieving high dependability and security. We show that using such clouds-of-clouds it is possible for services to continue to operate correctly despite dependability and security issues in a subset of the clouds. We show this approach with three cases: clouds-of-clouds for storage with the DepSky system; cloud-of-clouds for data processing with MapReduce; the execution of arbitrary services in clouds-of-clouds with the EBAWA algorithm. © 2014 Springer-Verlag Berlin Heidelberg.