BOSTON—As the April launch of Globus Toolkit 4.0 fast approaches, GlobusWorld opened here on Tuesday with the promise of bringing open-source, standards-based grid computing to a new level of usability, reliability and scalability and of bridging the gap between scientific and commercial grid.
“Were really open for business,” said Ian Foster, in his opening keynote. “The final release is in April, but the quality of the alpha is already better on the Web services front.”
The show follows close on the heels of the launch of two organizations that are devoted to commercialization of grid computing and which mirror the landscape of Linux—Globus Consortium, a vendor consortium made up of IBM, Hewlett-Packard Co., Intel Inc. and Sun Microsystems Inc. that is set up to be a nonprofit clearinghouse for matching requirements with resources to guide the development of open-source grid infrastructures, and Univa Corp., which is setting itself up as the Red Hat Inc. of grid, offering development and support of commercial grid solutions based on Globus software.
GT 4.0 is slated to bring the open-source grid toolkit up to date with the evolution of Web services standards. The Globus Toolkit is an open-standards building block for enterprise-level grid implementations that came out of the Globus Alliance, an open-source-focused organization at Argonne National Labs.
GT 3 was the first Web services iteration of the toolkit, and it suffered from growing pains that included poor documentation and licensing issues in its GridFTP transport component. GT 4 improves greatly on Web services implementation and also features a host of goodies, Foster said, including reliable file transfer, file-and-forget transfer, a Web services interface and integrated failure recovery.
“All these things are a journey,” Foster said in an interview with eWEEK.com following his keynote. Foster is the associate director of the Mathematics and Computer Science Division at Argonne National Laboratory and the Arthur Holly Compton Professor of Computer Science at the University of Chicago.
“Globus has been evolving from pre-Web services to Web services. GT2, which is still what the majority of Globus deployments are based on, is pretty robust and usable and scalable. The first Web services deployment in GT 3 was pretty immature software. It was hard to deploy, hard to manage. And scalability and performance were [flawed] in some places. In GT 4, its just much easier to install and manage.”
Scalability has also improved substantially in GT 4, Foster said. Processes that could scale to 100 services in prior versions can now scale to 1,000 services. Performance of security operations has also improved by an order of magnitude, with the implementation of Web services security standards.
What that means isnt so much fixing trapdoors or holes, Foster said, but instead pertains to usability around managing complex security policies that cover user access rights and monitoring.
What is the purpose of all this focus on grid? To solve the complexity of infrastructure that has cropped up as commercial enterprises and scientific bodies face applications mired in silos, with underutilized resources representing a huge waste of wads of cash.
The complexity issue is well understood, Foster said—however, the solution is not, as it encompasses myriad intertwined technologies, including Web services, utility computing, virtualization, data center automation and adaptive enterprise.
“The context for our work is the increasingly complex applications you encounter, thanks in large part to the emergence of the Internet and high-speed networks,” Foster said during his keynote.
“These applications may appear in the sciences. Alternatively they may come from industry,” with complex Web transactions sucking up large amounts of computing resources, he said.
For some time, people have been building applications to be dynamic in terms of resources they bring to bear on computing, Foster said. Often, however, they are implemented using proprietary technologies, often targeting a single, dedicated infrastructure as a basis for execution.
Building reliable, robust and secure applications is already difficult, Foster said. Doing it in a shared environment such as grid is even more difficult.
“Until recently, people avoided that and built overprovisioned silos for every [application] in the enterprise,” he said. “Globus is a recognition that its no longer feasible to build silos for every application we develop,” as it is a body committed to the idea of doing so in an open-source fashion.
Check out eWEEK.coms for the latest utility computing news, reviews and analysis.