EVALUATION OF DISTRIBUTED-SYSTEM TECHNOLOGIES FOR ALMA COMMON SOFTWARE
Catalogado desde la version PDF de la tesis.
Modern observatories develop distributed software systems to support their operations(e.g., data acquisition, hardware control and system monitoring). The complexity of thesesystems makes it difficult and even infeasible to build them from scratch so the softwaredevelopers rely on Off-The-Shelf (OTS) software, such as communication middleware, forthe development process. ALMA Common Software (ACS) is an open-source softwareframework for the development of distributed control systems that is based on CORBAmiddleware. Unfortunately the CORBA compliant middleware is struggling to keep up withthe increasingly demanding requirements of emergent observatories. This dissertationpropose an iterative solution, based on the PECA process created by the SoftwareEngineering Institute, for the evaluation and selection of communication middlewarealternatives for ACS, so the framework can be used by modern observatories, specificallyCherenkov Telescope Array (CTA). The process described in this work provides severaladvantages to the organizations working with OTS software. It reduces long-term cost andrisk of failure of the system, improves the knowledge about the system and its context,and allows the evaluation team to tailor the process to fit their needs and budget.Furthermore it also contributes to upgrade the reuse maturity of the organization and thetailorability of the process makes it applicable in many contexts such as the evaluation andselection of other key components of the system. Due to resource constraints this workonly provides a partial execution of the process, that aims to serve as a guide for furtherevaluation efforts. Even though no middleware candidate should be selected without afull implementation of the process, this first iteration describes the approach in detail andit generated several results that might be reused in future iterations such as a hierarchicalmodel of the multi-criteria decision-making problem, prioritization of the criteria, filteredpool of middleware candidates, data collection techniques recommendations, latencybenchmarking and coupling analysis of the most promising candidates.