Making scientific applications on the grid reliable through flexibility approaches borrowed from service compositions

The current trend in Service Oriented Computing (SOC) is to enable support for new delivery models of software and applications. These endeavours impose requirements on the resources and services used, on the way applications are created and on the QoS characteristics of the applications and the supporting infrastructure. Scientific applications on the other hand require improved robustness and reliability of the supporting Grid infrastructures where resources appear and disappear constantly. Enabling business model like Software as a Service (SaaS), Infrastructure as a Service (IaaS), and guaranteeing reliability of Grid infrastructures are requirements that both business and scientific application nowadays impose. The convergence of existing approaches from SOC and Grid Computing is therefore an obvious need. In this work we give an overview of the state-of-the-art of the overlapping research done in the area of SOC and Grid computing with respect to meeting the requirements of the applications in these two areas. We show that the requirements of business applications that already exploit service-oriented architectures (SOA) and the scientific application utilizing Grid infrastructures overlap. Due to the limited extent of cooperation between the two research communities the research results are either overlapping or diverging in spite of the similarities in requirements. Notably, some of the techniques developed in each area are needed but still missing in the other area and vice versa. We argue therefore that in order to enable an enterprise-strength service-oriented infrastructure one needs to combine and leverage the existing Grid and Service middleware in terms of architectures and implementations. We call such an infrastructure the Business Grid. Based on the Business Grid vision we focus in this work on presenting how reliability and robustness of the Business Grid can be improved by employing approaches for flexibility of service compositions. An overview and assessment of these approaches are presented together with recommendations for use. Based on the assumption that Grid services are Web services, these approaches can be utilized to improve the reliability of the scientific applications thus drawing on the advantages flexible workflows provide. This way we improve the robustness of scientific applications by making them flexible and hence improve the features of business applications that employ Grid resources and Grid service compositions to realize the SaaS, IaaS etc. delivery models.