Oct 17, 2005 |
The role of grids is on the cusp of a fundamental change. Specifically, a shift is taking place from grids being essentially IT infrastructure tools that tie together computing resources to grids playing a more central role in delivering and providing services to researchers. At the same time, the use of grids is also morphing from one that was exclusively the purview of academics and researchers to a state where organizations view grid technology as critical infrastructure needed to support their business operations.
The shift is duly noted by the Global Grid Forum (GGF), the organization that helps develop grid standards and specifications for grid interoperability. “We started about five years ago within the academic and research community,” says Mark Linesch, chairman of the GGF since September 2004. But he notes that membership has expanded to include members from many commercial organizations representing a wide range of industries.
“We have been successful in phase one — the introduction of grids,” says Linesch. “We focused on compute issues and bringing data together in a federated [type of] approach.”
This focus reflects the use of grids over the time period leading up to today. Notably, for about the last five years, grids have increasingly been used in the life sciences and other industries to harness CPU resources to apply more processing power to computational problems.
Examples of such use include Johnson & Johnson (J&J) expanding its R&D computational grid efforts from discrete departmental projects to into a companywide grid initiative to tap unused processing power of desktop PCs and some Linux clusters (see J&J Teams with United Devices for Global Grid, March 2005 Bio-IT World, page 40). When the J&J grid effort was announced earlier this year, David Neilson, senior director, J&J Pharmaceutical Research and Development IM (information management) said: “One business driver [for the grid] is to get maximum use of our current computers.”
Similarly, Novartis set up a grid that aggregated the unused CPU cycles of desktop PCs to run high-throughput docking and other life science applications. Within a few weeks of the initial deployment, 2,700 desktop systems were delivering about 5 teraFLOPS (5 trillion operations per second) of processing power and docking simulations run on the grid found an unknown CK2 inhibitor in the company’s library of compounds.
At the same time, efforts such as the National Science Foundation-funded TeraGrid and North Carolina Research and Education Network leverage both the computing and data-sharing capabilities of a grid.
With these deployments and others proving the value of grid technology, grids gained respectability as true corporate platforms for computing. And this led to the new phase in grids usage that Linesch refers to. “We need to look at grids in a broader distributed computing sense,” says Linesch. “Over the next five years, grids will allow sharing of resources and will connect users to disparate resources enabling [the delivery of] IT as a service.”
Laying the Groundwork
To bring grids into this next stage of use, the GGF has in essence a two-prong approach. The group has research and working groups that are developing grid specifications and best practices examples for grids in general and for industry-specific applications. In all cases, the GGF says its efforts are community-focused. “We are building out an international community of users and vendors,” says Linesch.
Some efforts cut across all industries. For instance, security and architectural issues such as how to support and use Web services technologies are important to pretty much anyone interested in grids today. And other issues, such as authentication, provisioning, and data management, that have to do with interoperability also cut across most applications.
In such areas, the GGF develops its own specifications and also reaches out to other organizations to leverage the work already done by other groups. “We realize we have to align with [existing] standards,” says Linesch. “For example, to develop a Web services resource framework, we worked with some of the other standards groups.” These other organizations included the World Wide Web Consortium (W3C) and the Organization for the Advancement of Structured Information Standards (OASIS).
The GGF also has liaisons with other groups including the Internet Engineering Task Force (IETF), Distributed Management Task Force (DMTF), and the Storage Networking Industry Association (SNIA). Linesch notes that the relationship with many of these standards bodies is “not only one-to-one but also includes roundtable talks about the issues.”
While many issues apply to all users, different industries have different requirements, so some of the GGF groups tackle industry-specific application issues. For instance, there is a life sciences grid group that looks at the issues related to the integration of IT with life sciences applications on a grid infrastructure. Within such groups, the focus is on developing use cases and best practices for a particular industry. For instance, a healthcare effort naturally could use all of the general grid architectural specifications, but it also focuses on how to meet HIPAA regulations in a grid environment.