When it comes to Desktop as a Service, the biggest hindrance has not been virtual desktop technology but rather Microsoft's licensing for Windows.
Before you dive into cloud-hosted desktops, make sure you understand the complex licensing rules that govern Microsoft Windows virtual desktops. Plus, keep your eye on a few other challenges you might encounter when deploying desktops via a cloud service.
Microsoft licensing for hosted virtual desktops
In order to access a virtual instance of the desktop operating system, Microsoft requires that endpoints have Virtual Desktop Access (VDA) rights. Microsoft VDA rights come with a Software Assurance for Windows subscription, or they can be purchased separately for devices that do not have or are not eligible for Software Assurance, such as certain types of thin clients.
Note that VDA rights are a device-centric licensing model in which each device accessing a virtual instance of Windows must be covered. Furthermore, the Companion Device License (CDL) was added with Windows 8 to "simplify" licensing from non-Windows endpoints. The Companion Device licensing is in addition to the Windows license and VDA rights.
Microsoft allows service providers and independent software vendors (ISVs) the option to license their software on a monthly basis by way of a Microsoft Services Provider License Agreement (SPLA). This is a pay-as-you-grow model as opposed to perpetual licensing models, where an enterprise pays a higher up-front cost for the license but then owns it forever. With SPLA licensing, the organization never owns the license; instead, it pays monthly at a much-reduced price compared with the perpetual license.
Unfortunately, there is currently no SPLA for Windows VDA, so users who want to use DaaS will need to purchase Windows VDA from Microsoft via their licensing agreements. As Microsoft declares, "Hosters need to ensure they isolate the hardware and other resources for each company. Any hardware running an instance of Microsoft software (OS or application) must be dedicated to a single customer."
You cannot have a single server that supports two different companies with users accessing Windows desktops on that server. This eliminates any ability to have multi-tenancy on the same hosts and increases the costs for the customer, as overall hardware use goes down. Underused hosts with capacity for additional virtual desktops cannot be pooled and used by other companies that require incrementally more resources. Therefore, dedicated pods of infrastructure must be used, and DaaS architecture for multi-tenancy is built around this Microsoft product use requirement. This is one reason why some DaaS providers require a minimum number of desktops; in small numbers, it's cost-prohibitive because of the dedicated physical infrastructure.
Some alternatives to delivering a Windows desktop avoid the SPLA, VDA and multi-tenancy constraints. Windows Server and Remote Desktop Session Host (Terminal Services) or technologies like Citrix XenApp, which use Remote Desktop Session Host, do not have the same restrictive licensing.
Desktop as a Service doesn't mean users have to connect to Windows 7 hosted virtual desktops on the back end. Some DaaS offerings include a Windows Remote Desktop Session Host branded to look and feel like Windows 7 but it executes on Windows Server instead.
Microsoft Windows Server and Remote Desktop Services (RDS) Client Access Licenses (CALs) are available on the Microsoft Service Provider License Agreement and do not prevent multiple customers from using the underlying hosting hardware. With this deployment model, multiple users can share a single Windows Server instance, or it can be deployed 1:1, with each user having his own Windows Server OS.
Other challenges of DaaS
More on DaaS challenges
Top five DaaS challenges
Pros and cons of cloud-hosted virtual desktops
Comparing in-house VDI and Desktop as a Service
Proximity to applications and data. The optimal place to deploy virtual desktops is as close as possible to back-end applications and user data. So, putting hosted virtual desktops across a WAN link from the back-end infrastructure and data usually decreases app responsiveness and results in a poor user experience. When DaaS is not located in the customer data center, performance can be worse than when the user desktop was on the LAN.
Operational management. Some DaaS providers do not provide support for the desktop operating system image or applications and rely on the customer to install, configure and optimize these components. In this case, much of the operational cost of supporting applications is still the responsibility of the customer, reducing any operational expenditure savings from using a DaaS system over an in-house virtual desktop setup.
Dig deeper on Virtual desktop infrastructure and architecture
Dan Brinkmann asks:
What do you think is the worst obstacle to implementing DaaS?
0 ResponsesJoin the Discussion