In addition, the Windows licensing rules aren't ideal for all companies, and can mean that DaaS isn't cost effective. "It's been a barrier," Forrester's Johnson says.
There are two potential setups. One is for a DaaS provider to use Windows Server with multiple user sessions running, meaning that multiple users are accessing their Windows desktops from a single instance of Windows Server.
Alternatively, a provider can run individual Windows desktop instances for each user. However, Microsoft requires that in these deployments, the hosting hardware is physically dedicated to a single customer. That ends up being very cost-ineffective, says Wes Miller, an analyst with Directions on Microsoft.
One potential downside to the Windows Server-based approach is that some applications may not run correctly in that environment. But it's not a problem with most apps, Miller says, so it's not enough of an issue to "justify the cost premium" of the Windows client implementation.
Also, while latency isn't currently an issue for the businesses interviewed for this story, it can be for some organizations. As a result, Johnson recommends that businesses use DaaS for employees who will almost always be on a high-speed network. This shortcoming, too, should improve, as communications protocols evolve to be more tolerant of network latency, he says.
"It's not a one-size-fits-all model," says IDC's Waldman. "It's not going to be appropriate for every user in your company."
Sign up for CIO Asia eNewsletters.