On June 26, the Global Grid Forum (GGF) and the Enterprise Grid Alliance (EGA) announced their merger to form the Open Grid Forum. Many will regard this as a very positive development. More importantly, this is a very necessary development. Why? At this juncture, Grid computing has limited ability to tolerate fragmentation in standards. In fact, Grid computing sorely needs to deliver definitive outcomes that really matter. Such outcomes depend upon focus in standards and elsewhere.

Having worked for Grid software vendor Platform Computing Inc. for about seven years, it took a little distance for me to appreciate all of this. To be perfectly blunt, Grid computing isn't quite as big a deal as many would like it to be. Consider the following data points.

Despite convergence efforts such as the Open Grid Services Architecture (OGSA), Web services continues to significantly eclipse Grid computing. At the highest levels, it's Web services that are synonymous with Service-Oriented Architecture (SOA), not OGSA. And for those who've waded into the reeds a little deeper, the evolution of the Open Grid Services Infrastructure (OGSI) into the Web Services Resource Framework (WS-RF) serves only to amplify this perception. Unfortunately for Grid computing, it doesn't stop there, as WS-RF has the potential to be engulfed by other WS-* standards or by simple RESTful approaches. This evolution in standards has wreaked havoc on the implementation front. For example, in moving from Versions 2 (protocol-specific implementation based on FTP, HTTP, LDAP, etc.) to 3 (introduction of Web services via OGSI) to 4 (refinement of previously introduced OGSI Web Services to WS-RF), the Open Source Globus Toolkit has undergone significant changes. When such changes break forward-compatibility in subsequent versions of the software, standards evolution becomes an impediment to adoption. Of course, a degree of standards evolution is inevitable. Unfortunately, however, the evolution that has taken place has not helped to elevate the profile of Grid computing relative to that of Web services. Extrapolating further suggests the possibility for collateral damage. As the gap between the promise and reality of the Semantic Web decreases, and Web services increasingly play a facilitative role, the Semantic Grid runs the risk of being reduced to a footnote.

Despite their validation of Grid computing, evidenced by their involvement in the GGF and endorsement of the Open Source Globus Toolkit, IBM markets more around virtualization. And virtualization, along with Web services, is also proving to be a big deal. Industry behemoths have checked in (e.g., EMC acquired VMware, Redmond developed the Microsoft Virtual Server), while the Open Source community enjoys Xen. Even Platform Computing makes the service orchestration that began a number of years ago with Platform Symphony more of a virtualization-than-grid play with more recent offerings (like Platform Virtual Machine Orchestrator). Whereas in-sourced virtualization is experiencing mass-market adoption, outsourced utilitization remains a mass-market nonstarter. Although the lackluster uptake of utilitization isn't specific to Grid computing, it's less than fortunate that "plugging into the Grid" is Grid computing's key message.

Grid computing startups still struggle to secure customers and, frankly, to survive. This is also in stark contrast to those companies that make Web services or virtualization their business. Moreover, a differentiated ecosystem continues to gel around Web services (just Google "Web services") and virtualization (like PlateSpin). This ecosystem features startups with compelling value propositions.

The highest-profile demonstrations of Grid computing run the risk of trivializing Grid computing. It may seem harsh to paint the well-intentioned World Community Grid as technologically trivial, but in terms of full disclosure, this is not the most sophisticated demonstration of Grid computing. Equally damaging are those clustered applications (like Oracle 10g) that masquerade as Grid-enabled. Taking such license serves only to confuse and dilute the very essence of Grid computing.

Applications can be effectively Grid-enabled by drawing on non-GGF, non-EGA or even Web services standards. Scali Manage 5 provides a compelling illustration by drawing on The Common Information Model (CIM), Web Based Enterprise Management (WBEM) and Eclipse. Whereas CIM and WBEM are standards that fall under the auspices of the Distributed Management Task Force (DMTF), Eclipse is an implementation framework and platform available from a consortium. By leveraging CIM, WBEM and Eclipse, a single instance of Scali Manage can manage multiple clusters in Grid computing contexts. Community-specific frameworks provide another illustration. In the case of the Earth System Modeling Framework (ESMF), a community of physical scientists is developing their own application-specific framework. This framework is required so that individual model components for the atmosphere, land, ocean and sea ice can interact across a network in real-time climate simulations. Although work on ESMF has been underway for some time, interest in Grid computing (and even Web services) is much more recent and underdeveloped.

I remain a Grid computing enthusiast, but as a realistic enthusiast, I believe that Grid computing sorely needs to deliver definitive outcomes that really matter.

That being the case, the recent creation of the Open Grid Forum is necessary, but is it sufficient?

About Ian Lumb

Over the past eight years, Ian Lumb had the good fortune to engage with customers and partners at the forefront of Grid computing. For all but one of those eight years, Lumb was employed by Platform Computing Inc. During the final year of the eight, Ian focused on Linux clusters while working for Scali Inc. About two months ago, Ian returned to Toronto's York University, where he is leading the networking team in the department of Computing and Network Services. Although he remains interested in Grid computing, his current research involves enhancing the semantic expressivity and richness of scientific data. Lumb's blog is available online at http://ianlumb.wordpress.com. The opinions expressed here are his own, and are not necesarily shared by his employer.

Last modified: Fri Jun 30 13:43:09 EDT 2006