Whittington notes that Florida Crystals' manufacturing systems remain on premises, not because they can't be run in the cloud but because there's still internal resistance to the move: "A number of manufacturing applications are still on-premises because the manufacturing folks have the old mindset," said Whittington.

Nevertheless, he's closed most of his data center, a move reflected in the fact that his peers would need 90 people to run the SAP applications that he's using, while he has three. Three people: a Basis manager--the SAP environment includes an ABAP language Basis technology platform for all apps--and two network managers constitute his SAP IT staff. His applications include most of SAP ERP, time-keeping, logistics, payroll, order management, invoice management, customer relationship management, and business intelligence. In addition to manufacturing, the main things still on premises are Microsoft Office applications, file, and print.

The movement of applications out of West Palm Beach occurred in several steps. Some applications were first placed on co-location servers 10 years ago with help from Adjoined Consulting, a specialist in packaged consumer goods IT, among other things. Rogers was a principal of Adjoined, which was eventually acquired by an outsourcing firm that was in turn acquired by Capgemini. Rogers left Capgemini with the former CTO of Adjoined, Kevin Reid, to form Virtustream.

"We wanted to bring the efficiency of full virtualization to the enterprise applications space," said Rogers over breakfast. By that he means the application server has to be correctly configured in a virtual machine to work with the application in its virtual machine, and the right connections established to the database. "You can kill yourself in a SAP environment ... If you don't have the app servers configured properly, it will bring you to your knees," he says.

After breakfast at the W last week, Whittington and Rogers presented a workshop at Structure 2011--"Best Practices for Enterprise Cloud Efficiencies"--on moving enterprise applications into the cloud. Their presentation was only lightly attended--it was competing with another event, lunch. One person who was there, Jack Shen, director/CTO of CBC Capital, listened carefully and asked questions at the end. I asked him afterward what he thought of the "enterprise cloud."

"We did something similar 12 years ago. We called it outsourcing. It's not really new in the enterprise space," he said.

Shen has a point, although outsourcers did not run client applications in multitenant setting on commodity servers. Part of what's different about cloud computing is that it rolls up a number of baseline advances--multi-core x86 servers, virtualization, and Internet-based networking--with a new way of distributing compute power and charging for it. Virtustream is doing that.

What caught Shen's eye was the usage-based pricing. Amazon Web Services has a price list for hourly usage for certain instance sizes. Virtustream has charges that are more closely based on actual usage, which can vary widely from one hour to another for ERP apps. It measures usage in "infrastructure" or "i units," with I/O weighing in heavily as a measure of actual activity. For example, Virtustream offers baseline pricing, when the customer knows how much CPU, RAM, and I/O activity he'll need. It has on-demand pricing for customers who want some elasticity built into their servers and they pay a higher-than-baseline price for it. Then Virtustream adds "beyond demand" pricing, where resources are guaranteed by the supplier regardless of demand level, with actual usage measured in "i units" that include I/O. It naturally carries a higher price.

"I don't get billed by VM. I get billed for what actually happens," said Whittington. Said Shen: "The I/O metric as a basis for billing, that's an innovation."

Whittington has taken a calculated risk that his SAP production applications will work just as well in a remote data center as they would in his data center. If he's wrong, and Florida Crystals suffers a business outage due to lack of accessibility to its systems, he'll be the goat. But if those 30% savings in IT operations pile up reliably year after year, he'll be a leading industry figure.

Whittington thinks he is getting reliable operations with secure data. He noted Amazon's service outage over the Easter Weekend but thinks Virtustream, which operates three of its own data centers, knows its "enterprise cloud" approach will be jeopardized if it ever experiences a similar temporary outage.

Rogers also claims he's aware of what a similar incident might do: "We stapled those accounts backwards to our foreheads," he said.

Charles Babcock is an editor-at-large for InformationWeek.

Data centers face increased resource demands and flat budgets. In this report, we show you steps you can take today to squeeze more from what you have, and also provide guidance on building a next-generation data center. Download it now.