There's a lot of debate these days over how appropriate virtual servers are for mission-critical applications. The concern is that adding an additional layer of software in the form of virtual machines will hurt these apps' performance.
But a study conducted by Principled Technologies on behalf of Stratus Technologies and NEC indicates that this doesn't necessarily need to happen if you have the right server architecture. The study shows that database performance scales linearly as the number of "virtual CPUs" increased, even though the amount of physical memory on the system stayed flat at 48GB.
Of course, the assumption here is that IT organizations will want to buy a new server to run mission-critical applications on top of virtual machines. There's no doubt that IT organizations can save money doing this; they just have to come up with the capital first.
But as Denny Lane, director of product management for Stratus, points out that with its servers, fault tolerance comes built in and prices start at $15,000. So no matter what happens with the performance of mission-critical applications on virtual machines, IT organizations can at least be certain that the application won't crash. And Stratus has put up a $50,000 guarantee to back that up.
When you think about the price of a server relative to the value of the mission-critical application, however, maybe it does make sense to spend a little now on new more efficient hardware than it does to try to rationalize existing servers running mission-critical applications. After all, what's going price for virtualization peace of mind these days?