What we're getting to now is known as a "private cloud" environment whereby the grid supports and manages the services delivery and integrity based on the ability to provision infrastructure instances and more efficiently balance services demand and services availability. This is not just a vision, but it is also not trivial to implement, and we're only in the early stages of cloud computing. Perhaps its best to try this out slowly.
For now, try and identify applications that may no longer be in large demand, but yet may still be supported with a full and expensive stack and proprietary platform. Can you offload that application or service from a full stack to a virtualized container in cheaper platforms? Are there, in essence, applications entering sun setting that are costing too much on their native ports, but which still need to be maintained? The grid can be a good place for new apps to emerge before demand is high, and as a "nursing home" for older applications as their demand wanes.
Over time, more applications and services in high and dynamic demand can be supported by the cloud. For now, your grid can support mission-critical applications designed for the grid, as well as non-mission critical applications and services modernized for the grid. This offers a financial payback while allowing you to gain more experience with grid, and using it for more diverse applications and services support.
This was first published in May 2008