Grid computing is democratic processing at its purest: From many equal parts comes one powerful response. A grid can be as small as a single computer focused on one problem, or as large as 1,000-plus processors clustered together to handle a wide array of requests. Basic CPUs running on a stripped-down operating system are managed by grid software and middleware that communicate with external requesting systems while coordinating the grid's internal processing power.

Our example assumes that after a thorough analysis, an organization such as a large university has decided to begin converting its data center to a grid computing infrastructure.

In the proper situation, a grid's clustering of basic servers into a web of processors can improve a data center's efficiency, processing power and reliability while reducing costs. For this project's rollout, cluster 16 basic processors, grouped into four nodes assigned to specific enterprise application and network requests. Linux will be your grid's operating system, and a suite of Oracle products will manage everything from "grid-ifying" data from highly customized application servers to figuring out which processor is open for work. Baseline created this calculator based on research, reporting and publicly available case studies and pricing; and interviews with Egenera, a Marlborough, Mass.-based maker of blade servers.

To see the details behind this Project Planner and fill in your own estimates, click on the "Get the Tool" icon above and download the interactive worksheet.