DaaS vs. IaaS for Desktops
Let's continue the cloud conversaton that I brought up in last week's blog, but this time on another topic that has garnered steam in the last few weeks among my customers: Desktop as a Service. Customers are now asking, why DaaS instead of VDI?
I don't want to turn this blog into a comparison between them and this fight has been discussed to death in other forums. Still, I'd like to highlight a few things that DaaS needs before it is a viable alternative to VDI. The biggest hurdle is Microsoft licensing. At the moment, the company doesn't have a Service Provider License Agreement for its desktop operating system products and that means customers have to provide their own Microsoft licensing to their DaaS provider. I have a problem with that -- without one, it gets very complicated, even more so than VDI Besides, it then is no longer provided in an "as a service" model.
Here's another hurdle: DaaS providers are delivering Windows Remote Desktop Session Host desktops and accessorizing them with a Windows 7 theme, and that presents its own set of challenges with apps and other considerations. There is also the concern with data ownership and compliance. Most important, DaaS would be limited to SaaS applications or Windows applications that are self-sufficient, meaning they don't need access to the corporate data or back-end databases. These are just very quick nuggets of some show-stoppers that I see at the moment.
That being said, I believe DaaS will eventually surpass VDI adoption once some of the obstacles are addressed. However, one alternative I have not heard many discuss yet is IaaS for desktops. I believe it will be the most popular and cost-effective solution. With DaaS, you lose a lot of control over customization of desktops, and management flexibility, which is understandable. Service providers cannot have an unlimited number of change requests, as that becomes a nightmare to manage and support.
IaaS for desktops brings the cost savings that IT is looking for, liberates IT from the hassle of managing the infrastructure, and it's customizable to meet the requirements of an organization at an SLA level. Instead of SP providing the desktop, they provide customization at an IOPS level, compute level, etc. All IT has to do is deploy and manage their VDI environment atop an IaaS customized for desktop workloads. This scenario makes a lot more sense for enterprises to acquire their own Microsoft licensing, as they are simply swapping out a physical infrastructure in their datacenter for a cloud-based IaaS designed for desktop workloads. They still have to do everything else.
It's a scenario that can give IT that warm and fuzzy feeling that they are still in control but are leveraging the cost savings and scale of the cloud. You still have some challenges to overcome in that if your entire data and applications are not in the cloud, preferably on the same IaaS with a different SLA, you still have performance challenges. But if you entertained the idea that I raised last week that in five years most enterprises will be on an IaaS platform instead of their own datacenter infrastructure, then this approach makes a lot of sense.
Put all your technical reservations aside for a second when you think of DaaS. We understand the separation of data from compute will not give optimal performance, but then again, I'm assuming in five years IaaS will be the norm. With those caveats, do you see a DaaS or an IaaS for desktops scenario being more relevant? Share your comments here.
Posted by Elias Khnaser on 05/06/2013 at 12:49 PM