Fault-aware management protocols for multi-component applications

Nowadays, applications are composed by multiple heterogeneous components, whose management must be suitably coordinated by taking into account inter-component dependencies and potential failures. In this paper, we first present fault-aware management protocols, which allow to model the management behaviour of application components, and we then illustrate how such protocols can be composed to analyse and automate the overall management of a multi-component application. We also show how to recover applications that got stuck because a fault was not handled properly, or because a component is behaving differently than expected. To illustrate the feasibility of our approach, we present BARREL, a proof-of-concept application that permits editing and analysing fault-aware management protocols in multi-component applications. We also discuss the usefulness of BARREL by showing how it was fruitfully exploited it in a concrete case study and in a controlled experiment. (C) 2018 Elsevier Inc. All rights reserved.