Software licensing has never been simple, but the issue is becoming particularly thorny as IT managers deploy new data center technologies such as server virtualization, grid computing and multicore processors that stretch the limits of traditional per-CPU pricing.

These shifts in data center architecture bring more flexibility and efficiency to hardware platforms, but they also are likely to result in higher costs - as much as double the cost , in some cases - for software that is priced on a per CPU basis, analysts say.

That is causing many users to take a cautious attitude when deploying these new technologies. Eric Kuzmack, IT architect at newspaper conglomerate Gannett in Silver Spring, Md., says he is waiting to see how software vendors decide to handle the licensing issues

"A lot of it really comes down to whether vendors are looking at dual core as a way to increase their revenues at our expense, or as a way for us to be able to better utilize their software," he says.

That's what makes the dual-core pricing a sticky issue. It's not clear what kind of performance improvement end users will see from dual-core systems vs. single-core systems.

"The problem is dual core will not necessarily make applications twice as fast. From what we've been reading and hearing, it's nowhere close to a doubling," Kuzmack says. "Depending on what the application is, if the application isn't written to take advantage of multiple processors to begin with - or if it isn't written very well - then adding more cores won't give you the bang for the buck that the vendors would like to think it would."

Combine the industry's move to multicore processors with another trend toward server virtualization and grid computing and IT managers could be facing huge bills when it comes to their software licenses, analysts say.

"With technologies such as multicore, virtualization, grid and on-demand, the per CPU model is falling over," says William Fellows, principal analyst at research firm The 451 Group.

The issue is that the traditional one application/one server scenario is being replaced by more flexible environments where software is no longer tied to hardware and CPU usage rises and falls according to demands. As a result, grid or virtualized environments may require end users to pay for the total number of CPUs potentially available to an application, while only a fraction of those CPUs are used most of the time.