Testing distributed middleware frameworks requires creating many different configuration combinations to exercise various component interactions and resource allocations. Manual creation of these test configurations is error-prone, time-consuming, and cannot scale to cover all meaningful combinations. The innovation lies in applying model-driven techniques to automatically derive test cases from system models, improving test coverage and reducing manual testing effort.
This paper develops a model-driven tool for automated system-level testing of distributed middleware frameworks like XDAQ. The approach uses Model Integrated Computing (MIC) principles to automatically generate test configurations and deployment scenarios from high-level system specifications. The tool combines behavioral modeling with resource and deployment specifications to create comprehensive test suites that exercise different system configurations without manual effort.
The paper presents a testing tool infrastructure that uses Model Integrated Computing to generate test configuration models from domain-specific specifications. The tool successfully generates XML configuration files and test scripts for the XDAQ middleware framework used in high-energy physics experiments. The approach demonstrates how model-driven techniques can automate test generation for complex distributed systems.
@inproceedings{Nordstrom2006a,
author = {Nordstrom}, S. and Dubey, Abhishek and {Keskinpala}, T. and {Neema}, S. and {Bapty}, T.},
booktitle = {Third IEEE International Workshop on Engineering of Autonomic Autonomous Systems (EASE'06)},
title = {GHOST: Guided Healing and Optimization Search Technique for Healing Large-Scale Embedded Systems},
year = {2006},
month = {mar},
pages = {54-60},
abstract = {Reflex and healing architectures have been shown to provide adequate user-defined initial failure mitigation behaviors in the presence of system faults. What is lacking, however, is a user-guided means of healing the system after the initial reflexes have been enacted. This process should be autonomic in the sense that new system configurations can be achieved by defining a priori only a small set of criteria to which the healed system should conform. What follows is an explanation of this technique for guided healing which allows system designers to direct the healing process from a higher level in such a way that the resulting system configurations satisfy their particular needs. A brief example outlining the application of this approach is given},
category = {conference},
contribution = {minor},
doi = {10.1109/EASE.2006.8},
file = {:Nordstrom2006a-Ghost_Guided_Healing_and_Optimization_Search_Techniques.pdf:PDF},
issn = {2168-1872},
keywords = {model-driven testing, middleware testing, test configuration, domain-specific modeling, distributed systems, test generation, Model Integrated Computing},
month_numeric = {3}
}