TACC Releases Beta Version of GridPort 3.0

A portal is a window through which a user may find and exercise the many features of a fully capable computational environment. Through a portal built with GP3, a scientist or engineer can, with a single login, use computing resources, storage systems, databases, remotely controllable scientific instruments, distributed networks of sensors, and visualization facilities. "GP3 is designed to make it easy to build specialized user interfaces and to construct composite tasks such as workflows. Our focus is on portals for groups of scientists or engineers who access computational resources of all kinds," said Mary Thomas, GridPort PI and Grid research scientist at TACC. "We are very excited about being able to support a new range of capabilities with this version." "In building GP3, we have completely revamped our widely used Grid Portal Toolkit to take advantage of evolving technologies and user requirements," said Maytal Dahan, GridPort project manager and software developer at TACC. "New features enable users to launch and control multiple jobs on multiple resources and to determine the sequencing of such jobs in whatever order is desired." With GP3, Dahan noted, Grid users need no longer climb a steep learning curve. The complexities ordinarily involved in using Grid software to work on disparate systems are hidden from the user, who can concentrate on the scientific experiment or problem without worrying about the details of the underlying resources. Obtaining the GP3 Beta Release The GP3 software and a demonstration portal built with GP3 are available for download from the GridPort website (http://www.gridport.net). Interested users can also find more information at the website, subscribe to informational mailing lists, and contact the GridPort team directly. History and Design In use since 1999, the portal-building toolkit was developed by researchers at TACC and the San Diego Supercomputer Center. It made possible the well known "HotPage" access to the supercomputing resources of the Partnerships for Advanced Computational Infrastructure. The original toolkit and the second version (GP2) were written in the Perl language and included user authentication; job submission; file and data management; process/command execution capabilities; and account and session management tools. These services were aggregated and made available through Web-page portal interfaces that could be customized for specialized groups of computational scientists. GP3 is written in Java, which has allowed the addition of features that use new Web- and Grid-based tools. The design "represents a significant shift away from the ultralight-weight approach of the original toolkit," Dahan said "But we wrote it with two major environmental changes in mind. One was that the increase in the numbers and kinds of disparate resources has stimulated the design of a layer of web services, which are being made accessible throughout the computational environment: compute platforms, storage area networks, large-scale instruments like telescopes, and networks of tiny environmental sensors. These services need to be selectable and integrable at single access points for specific needs. The second was that the increase in the number and diversity of virtual organizations (collaborations and multi-institutional groups) employing such resources made a compelling case for increasing GridPort's capacities." GP3 thus requires more server-side resources, "but this is more than balanced by increased capability and scalability," said Tomislav Urban, senior software developer at TACC and chief software architect for GridPort. A key enhancement is a service called the Grid Portal Information Repository (GPIR). "GPIR ends the transient nature of GridPort operations," he explains. It can store static data about resources –such as machine descriptions and types--or dynamic data such as load, job, queue, and node information. It is thus a repository for information about resources and user groups. The GP3 software has been designed to be compatible with standards for Web services that are accelerating the transition of Grid computing to a Web services-based architecture"GP3 also supports the easy integration of external services," Urban said, such as the Globus Toolkit v. 3 and job sequencing using the Community Services Framework (CSF) developed by Platform Computing. These services are available in the demonstration portal that comes with the downloaded GP3 software. "The aim is to provide a single, integrated, high-level Applications Programmer Interface spanning a comprehensive set of low-level Grid services," Urban said.