Consider these cautionary tales from reader Steve Ruger, an independent project manager in North Carolina. One client, a large computer-chip manufacturer, was replacing its ERP (enterprise resource planning) system, and the first round of infrastructure sizing--done after the budget was prepared--indicated that the original $22 million budget needed to be closer to $30 million.
"As I walked in there, the CIO's career was flashing before his eyes," Ruger says. "He was going to have to ask for an additional $8 million." Fortunately, after going through the vendor's sizing algorithms, which didn't take into account usage patterns, Ruger was able to determine that the sizing for the network was more like $21.5 million. Lesson: Get input on sizing from more than one source.
Ruger also saw a horror show at an international train manufacturer, where data load was calculated on a small system and then predicted in a linear fashion on the huge production system. The team couldn't do a system conversion simply because "that big of a slug of data across at one time would take a week, and we couldn't shut down both systems for a week," according to Ruger. It ended up taking the team a month to do a parallel conversion.
Lesson: To successfully create a project budget, be careful how you measure needs. Don't think of scalability as being necessarily linear. And by all means, be reasonably sure of your sizing before creating the budget.
But what do you do about infrastructure upgrades--you know, the stuff that no one department owns and doesn't have a definite return, but that you believe would make a big difference for the entire company?
Unfortunately, unless your company is flush with cash, the only responsible thing to do is not upgrade unless you can identify a business benefit. There is, after all, a fine line between a benefit and a rationalization.