Portlet-based Grid portals have become a crucial part of the international distributed computing infrastructure to support computational science ("cyberinfrastructure") provide component-based problem solving environments for scientists. Although Web portals are intended to provide user-friendly environments with easy-to-use interfaces, the development of portals and their portlet components is time consuming. We aim to address this problem by providing reusable components for rapid portlet development. Our approach, Grid Tag Libraries and Beans (GTLAB), encapsulates common Grid operations with reusable XML tags, providing a dramatically simplified programming interface to common cyberinfrastructure services. GTLAB also provides a way for creating composite tasks that models the requirements of computational science portals. In addition to standard Grid job submission and remote file operation tags, we also provide management and monitoring capabilities for Grid tasks. This system can persistently store job metadata, which results in a permanent storage for archiving and reference.

In this dissertation, we have studied and observed two distinct science gateways as use cases. First, the QuakeSim portal is a problem solving environment to develop a solid Earth science framework for modeling and understanding earthquakes. In this study, we have proposed an evolutionary approach to allow TeraGrid usage in addition to clusters for QuakeSim portal. Second, VLab is a Grid and Web Service-based system for enabling distributed and collaborative computational chemistry and material science applications for the study of planetary materials. The requirements of VLab include job preparation and submission, job monitoring, data storage and analysis.

Although GTLAB provides support for simple Grid workflows, we must also investigate the problem of integration with existing workflow systems. We have thus extended GTLAB to support the widely used Condor DAGMan and Taverna workflow systems. These extended tags demonstrate that large workflows can be integrated within Grid portlets without burdening the developers.