In November we interviewed several key analysts in the Grid space. As a follow-up we posed the following question to our past contributors. "What do you predict will be the leading story (or stories) for Grid in 2007?" This feature is a summary of their responses.

If the past year's Grid news can be described in terms of two main trends-the consolidation of the Grid community, and the concurrent proliferation of Grid-derived technologies-then in the coming year, we expect to see more of the same. Except different.

Come again? The formation of the OGF last spring was definitely the most significant indication in years that the Grid community is coming together for everyone's mutual benefit. When the business and research communities came together in the OGF to promote Grid technology on a global scale, the move grabbed headlines and set the blogosphere atwitter. But a quieter revolution has been gaining momentum on a local level, as more and more individual grid projects achieve interoperability and pool their resources for larger purposes.

Indeed, the multipurpose conglomerates like TeraGrid and Open Science Grid (OSG) are taking over from the first generation of so-called "community grids," which catered exclusively to the served the needs of specialized communities, like petrochemical explorers or high-energy physicists. The great news is that these giants are not prone to the dreaded "forking." Instead of competing, they're cooperating: TeraGrid has partnered with the Distributed European Infrastructure for Supercomputing Applications (DEISA), Enabling Grids for E-SciencE (EGEE) is working with the OSG, and the e-Infrastructure Reflection Group (e-IRG) of the European Commission is helping to coordinate Europe's national grid initiatives. Some of the more advanced European grids are drawing upon upwards of seventy distinct institutional partners!

We've said it before, but we'll say it again: bigger pools make for better parties. Of course, we hope the Globus Toolkit will be enough to compel pooling of resources, but where convergence to a single middleware stack is impossible, there's still hope. The OGF's Grid Interoperation Now (GIN) group is facilitating middleware interoperability, and Germany's D-Grid Initiative has demonstrated that supporting Globus, gLite, and Unicore can indeed be done.

On the enterprise side of things, standards are improving, particularly for domain specific workflows. Accordingly, as it becomes increasingly feasible to compose a variety of compute and data services to provide particular functionalities, we may well see a shift away from implicit workflows toward smoother domain specific jobs in the coming year.

Likewise, interest in traditional HPC Grid deployments is giving way to more and more demand for throughput compute grid installations. This year, for instance, Web leaders like Google, Amazon and Ebay won the attention of the Grid community by horizontally scaling their throughput computing, and experimentally migrating transactional applications on Grids of commodity machines. This is the kind of forward thinking that could eventually drive wider Grid adoption, especially within those industries that have little need for HPC resources. Tried and true HPC customers like financial firms are not going anywhere soon, but in order to reach even more industries it seems clear that Grid must integrate into IT architecture in new and powerful ways.

Of course, one of the most closely watched stories of 2007 will be, no doubt, just this: how do enterprises use the latest wave of Grid-derived technology? The concepts of virtualization and SOA based applications will obviously inform the next wave of data center design and IT architecture. The question is how...

close window