[communities] GGF Proposal Submission

[david.manset at cern.ch]

proposers_name: David Manset 
 
affiliation: CERN / University of the West of England (UWE) 

email: david.manset at cern.ch 

proposed_title: A Model Driven Approach for Grid Applications Engineering 

session_type: individual presentation 

proposed_duration: 30 mins 

target_audience: technical experts, researchers 

num_attendees: 30 

abstract: This paper discusses the concept of model-driven software engineering applied to the Grid-based application domain.  As an extension to this concept, the approach described here, attempts to combine both formal architecture-centric and model-driven paradigms. It is a commonly recognized statement that Grid systems have seldom been designed using formal techniques although from past experience such techniques have shown advantages. This paper advocates a formal engineering approach to Grid system developments in an effort to contribute to the rigorous development of Grids software architectures. This approach addresses quality of service and cross-platform developments by applying the model-driven paradigm to a formal architecture-centric engineering method. This combination benefits from a formal semantic description power in addition to model-based transformations. The result of such a novel combined concept promotes the re-use of design models and eases developme!
nts in Grid computing. 

synopsis: The session\'s main objective is to assess the relevance of Model Driven Engineering (MDE) for Grid computing. Model driven engineering can be applied efficiently to different areas in Grids including Grid applications development, data organisation and treatment as well as job scheduling/workflow descriptions. 
This session would provide an example of how the model driven concept can be applied to particular development issues. As a case study, the proposed research investigation will give concrete measures of the applicability of MDE in the European MammoGrid project (www.mammogrid.com).
Indeed, one of the major issues in today’s Grid engineering is that it often follows code-driven approaches. Although it has been proven from past experience that using structured engineering methods would ease the development process of any computing system and would reduce complexity, the hype of Grid computing is still encouraging ‘brute-force’ coding and consequently a rather unstructured engineering process. This always leads to a loss of performance, interoperability problems and generally ends in very complex systems that only dedicated and expert developers can manage. As a direct consequence the resulting source code is neither re-usable nor does it promote dynamic adaptation facilities as it should, since it is a representation of the Service Oriented Architecture (SOA) paradigm. As it is directly extracted from the definition of the Grid concept, Grid design should ensure cross-platform interoperability by providing ways to re-use concretely systems in a heterogen!
eous context. Having no guidelines or rules in the design of a Grid-based application is a paradox since there are many existing architectural approaches for distributed computing applications which could ease the engineering process, could enable rigorous engineering methods and could promote the re-use of software components in future Grid developments. 

MDE appears as a concrete and relevant solution reconciling the wide variety of development processes and techniques, which could be helpful to Grids. MDE is becoming mature for taking over the next step in software engineering. 

tech_requirements:  

prereq_participants: Model Driven Software Engineering, Grid computing 

advertise_suggestion: