August 18, 2003, 11:33 AM — As much as it is a mistake to assume that negativity is a sign of intelligence, it is also a mistake to succumb to blind optimism. And so we are raising a glass to the critics, cranks, and curmudgeons who alert us to the stumbling blocks and help us perfect the vision.
Following are five broad categories of potholes on the road to on-demand nirvana.
1. There's no there there
In many, if not most, ways, utility computing is still vision, still vapor, an idea that is not quite reality. That's the nature of emerging technologies, and it is especially true of utility computing, which is exceptionally wide reaching and is itself comprised of emerging technologies, including blade servers, autonomic computing, virtualization, Web services, and grid computing.
2. Money... it's a drag
No two ways about it, organizations will have to do their math when considering when and whether to jump into utility computing. In a report titled Distributed Computing Economics, Jim Gray of Microsoft Research finds that "'On Demand' computing is only economical for very CPU-intensive (100,000 instructions per byte or a CPU-day per gigabyte of network traffic) applications." Futhermore, even if an organization can clearly benefit from utility computing, existing architectural barriers may be insurmountable -- at least in the near term.
3. Security, security, security
Any IT professional worth her salt is right to get a little fidgety around the idea of sharing computing resources with other organizations. Security is a critical element of any organization's IT strategy, and it is not a topic on which the visioneers wax poetic. Bottom line: make security a priority when selecting a vendor.