Overcapacity issues in data centers can affect cooling, compromise floor space and elevate power usage. But as this case study shows, a return on investment (ROI) calculator
When server performance at OpenSolutions Inc. began to suffer because of overcapacity and antiquated equipment, the company had to make a choice.
"We had a lot of very old equipment," said Ryan Marsee, director of corporate networking for OpenSolutions, a financial services provider in Glastonbury, Conn. "The only way to get fix it was to buy all new equipment or go to virtualization."
The virtualization ROI calculator as a selling point
The already-taxed OpenSolutions data center needed the ability to grow following a recent acquisition, but physical expansion wasn't an option.
"We were at 100% capacity from a cooling perspective," Marsee said. "Even if we resolved this, we had no floor space left."
Using his company's proprietary ROI calculator, Marsee worked the numbers.
"Fixing the cooling issue would cost around $250,000," he said.
And fixing only the cooling issue meant the company would keep things status quo for the equipment in the data center -- a step above putting a Band-Aid on the problem. Ultimately, that solution didn't allow any room for growth. Instead, Marsee considered eliminating some of his energy-inefficient monster servers. Although management rejected the idea of virtualizing servers in the past, the $250,000 quote on a new cooling system made virtualization seem that much more attractive to the CIO.
Factors the virtualization ROI calculator included were the need to increase cooling, the need to decrease the cost of a per-server deployment -- logical or physical -- and the desire to reduce the center's 15:1 compression ratio.
"Based on this, we were able to realize about a seven-month ROI on the initial virtualization project," Marsee said.
Future virtualization ROI calculator plans
Using the virtualization ROI calculator from the onset gave the company a good idea of what to expect in virtualizing the data center. It also shined a light on some known inefficiencies in the corporate networking infrastructure that had to be fixed anyway.
This was first published in April 2010