MuScADeL: A Deployment DSL Based on a Multiscale Characterization Framework

Abstract

With the Internet of Things (IoT) paradigm, ambient systems move from locally distributed systems to Internet distributed systems. These systems become huge in term of number of devices and imply high heterogeneity (e.g., of devices, of networks). They are continuously evolving with appearing and disappearing devices at runtime. The inner complexity of these systems, called multiscale systems, requires autonomic deployment middleware. Such middleware should deploy components where and when necessary, and adapt the architecture of the deployed systems considering the different scales of the systems. In this paper, we define MuScADeL, a domainspecific language (DSL) dedicated to multiscale and autonomic software deployment. MuScADeL allows designers to abstractly define deployment properties without exact knowledge of the devices and networks the system will be deployed on. This DSL is based on a scale-awareness framework, which helps designers to characterize the multiscale nature of a system from several viewpoints such as device, network, administration and geography. With MuScADeL, deployment designers may express multiscale properties of systems to deploy. MuScADeL is a building block for deployment middleware that targets multiscale distributed systems. We illustrate the possibilities of MuScADeL through a smart transport scenario.

Predictable Efficiency for Reconfiguration of Service-Oriented Systems with Concerto

“…With the increasing complexity of both distributed software systems and distributed infrastructures, such reconfiguration procedures have to be automated, and may ideally autonomously react to environmental events. Autonomous reconfiguration is usually modelled with a control loop, such as MAPE-K [1], [2], [3]. It consists in the following four steps: (M) Monitoring that returns information about the current configuration of the system as well as events of interest; (A) Analysis that, given the information gathered by (M), chooses a new target configuration for the system; (P) Planning that determines a reconfiguration plan to go from the current configuration to the new one; and (E) that executes the plan returned by (P).…”