Identifying Failure Causalities in Multi-component Applications

Understanding and resolving failure causalities in modern enterprise applications is one of the main challenges daily faced by application administrators. Such applications indeed integrate multiple heterogeneous components, and identifying which components are causing the failure of which other components requires to delve through distributed application logs. In this work-in-progress paper, we present our idea of devising an analysis approach based on management protocols, a fault-aware compositional modelling for the management of multi-component applications. We discuss how they can be used to identify causalities of failures in multi-component applications, and to design countermeasures to avoid (or, at least, limit) failure propagation.