An October 2003 report by the 451 Group was entitled Grids 2004: From Rocket Science to Business Service. In the Key Findings Section we can find several optimistic predictions. For example: Grid computing developments will result in commercially viable, mainframe-like performance and manageability across distributed systems within the next 12 months. The report listed financial services, life sciences and manufacturing vertical markets as the early adopters of grid computing. At the same time the report warned that grid computing can stumble if vendors do not provide sufficient security and authentication.
In 2006 we can add other observations. The most important of them is grid computing has become a key technology for products and services of several significant vendors including IBM, HP and Sun. In addition grid computing has captured imagination of most IT professionals from academic and government institutions. Every year from 2000 we have witnessed many respectable conferences discussing, advocating and pushing grid computing. Despite this formidable effort and related expense we cant say in 2006 that grid computing has succeeded and has become a common utility computing technology. A simple explanation is related to human factors and the existing computing culture. After several decades of computing based on hardware ownership and total control of resources and services grid computing is a major paradigm discontinuity. Those who need very large amounts of computing or need to collaborate with geographically dispersed partners have used grid computing successfully. Others still wait. More complex explanations must take into account other factors such as: business data and software security, benchmarking difficulties, IT staff retraining and the expenses of the paradigm shift implementation.
In conclusion what we need now is not more vendor business strategy declarations and scientific conferences pushing grid computing but a careful and open analysis of the stumbling blocks on the way to Enterprise Grid Computing. This analysis could be done by several leading vendors together with potential commercial users who are not interested in early adoptions but safe, economically justified real life applications of grid computing. The commercial user companies participating in the analysis should not be only large rich businesses that are afraid of missing the boat and can afford some experimentation. The analysis should include small and medium companies that need large amounts of computing for their real-life business functions.
Janusz S. Kowalik