We consider the impact of different communication architectures on the performability (performance + availability) of cluster-based servers. In particular, we use a combination of ...
We study several transparent techniques for scaling dynamic content web sites, and we evaluate their relative impact when used in combination. Full transparency implies strong dat...
This paper presents an experimental study on the emulation of software faults by fault injection. In a first experiment, a set of real software faults has been compared with fault...
We propose a two-phase methodology for quantifying the performability (performance and availability) of cluster-based Internet services. In the first phase, evaluators use a fault...