Hierarchical self-optimization of SaaS applications in clouds

Abstract

This chapter introduces a framework and a methodology to manage a SaaS application on top of a PaaS infrastructure. This framework utilizes PaaS policy sets to implement the SaaS provider's elasticity policy for its application server tier. Adaptation is based on strategy-trees, which allow for systematic capture, representation and reasoning about adaptation variability, based on hierarchically organizing different levels of temporal granularity. Thus, a strategy-tree is utilized at the SaaS layer to actively guide policy set selection at runtime in order to maintain alignment with the SaaS provider's business objectives. This way, the SaaS provider's attitudes and preferences reflecting their general business needs are incorporated into the adaptation mechanism in an organized and accessible manner. Results from an experiment conducted on a real cloud are presented in support of this approach. © 2013 Springer-Verlag.