Dependability in Software Component Deployment

Abstract

The deployment of distributed systems is extremely complex. Indeed, software deployment faces significant problems of success and safety. These problems are due to the lack of a unified model, a global architecture for deployment and an explicit dependency representation among the target system and application components. Therefore, we advocate moving from an ad hoc approach to a structured, systematic and safe approach. In this paper, we propose a generic model and an architecture for a dependable deployment.

A taxonomy of software architecture-based reliability efforts

“…Few reliable activity works aim to incorporate architecturebased reliability guarantees in the operational activities (deployment [3], configuration [13], and upgrading [8]). Reliable deployment ensures the appropriate functionalities and the correctness of the system after the deployment activity.…”

“…The main goals of the architecture-based reliability efforts are fault tolerance [14,30,31], reliability analysis [6,15,34], and reliable design and operational activities [1,3,8]. Fault tolerance is achieved by tailoring the fundamental and traditional fault tolerance approaches to suit the software architectural styles, mainly by introducing software redundancy to system architecture levels.…”

SALSA: A Framework for Dynamic Configuration of Cloud Services

“…Krzysztof et al in [18] introduce an approach of incorporating Domain Specific Languages(DSL) into the process of developing and deploying applications. Meriem et al in [19] propose a deployment model to manage the dependencies for software components for the deployment. Binz et al in [12] show approach to use TOSCA to resolve the dependencies of the cloud service.…”

“…The heterogeneous service unit types and relationship types in multi-cloud environments bring challenges in resolving dependencies and require a base representational model to deal with the problem. Studies in [12], [13], [19], [29] proposed service unit models for resolving dependencies for cloud services. These models focus on relationships between service units at the deployment time but not during runtime.…”