With Amazon’s recent announcement of the Workspaces offering and VMware’s Horizon DaaS offering, customers have started to inquire about the relevance and reality of such a solution.  As a Sales Engineer, I address these questions from customers on a daily basis. While I believe DaaS is here to stay and might be a perfect fit for some (especially in the SMB space), I don’t think it is a solution for a majority of the enterprise customers out there today. I thought I would post my views why I believe DaaS is not the Silver Bullet.

(I want to be clear that the views expressed here are my own.)

Cost:

While upfront, DaaS seems like a cost effective solution (Amazon Workspaces and VMware Horizon DaaS offering starts at $35/month), it offers you a very limited environment. When you size something similar to what end users are used to having as their corporate devices (beefy processor, 4G Memory, 100 GB HD), all of a sudden that cost increases (upwards of $65).  Now how about your power users?  You get the idea.  This does not include applications.  So when you factor everything and the costs associated with maintaining some multi-tiered applications on-premises, I would argue that the cost of DaaS could be a wash or higher when compared to an on premises solution.  I am sure over time the cost of DaaS will come down significantly and the application architectures will change and will suit the DaaS model.  But as of today, cost is not a reason to move to a DaaS model for most enterprise customers.

Uptime, Reliability & SLAs:

When considering DaaS, keep in mind that the customer has no control over the backend infrastructure and is totally at the mercy of the service provider when it comes to reliability, DR and uptime.  If a large service provider such as Amazon can have outages, and be down for extended periods of time, you have to seriously question the reliability of DaaS.  One way to mitigate risk is to define a robust Service Level Agreement (SLA), but this can be challenging due to lack of flexibility by the service provider.  In an on-premises implementation, the customer can architect a fully redundant and fault tolerant solution that the customer can fully control.

Performance/Complexity:

For customers that want to provide a desktop with just Office productivity applications, DaaS could be a viable option.  Small businesses would be an example where such a solution would be a good fit.  However, for Enterprises, the application portfolio is much larger and consists of a variety of tiered client-server applications.  Most enterprises are hesitant to move the backend data into the cloud for various reasons.  So in order to make the DaaS solution work for enterprise customers some form of a VPN tunnel is required that bridges the cloud infrastructure with the customer on premises infrastructure.  The customer has to manage infrastructure on-premises and in the cloud.  This adds complexity, makes troubleshooting harder, could lead to performance issues, and could make DR and SLA agreements extremely complicated.  Also, in a scenario where users connect from different parts of the world and have data living locally, there could be various challenges such as:

  • Performance Issues:  If the DaaS provider does not have a presence across the globe, there could be serious performance issues when accessing the desktop and more importantly, the applications.  In a lot of cases, businesses or clients might control where the data is stored.  If certain application data is stored locally in a different part of the world, there could be challenges in having the data synchronized at all times and could also lead to performance issues.  In a traditional VDI model, desktops move closer to the applications in the data center which leads to better performance whereas with DaaS, desktops move further away from applications which is bound to impact the application performance and user perception.
  • Connectivity and Bandwidth Issues:  One of the assumptions when going down the DaaS route is that the user is always connected. The challenge is that if there are any connectivity issues either at the service provider or the user, the user loses access to their desktop.  For a large enterprise that accesses thousands of desktops in the cloud, there could be a significant bandwidth requirement that could drive up costs.
  • Maintaining Infrastructure (on-premises and in the cloud):  In a lot of cases, data and application infrastructure will reside in the customer’s private cloud with the DaaS infrastructure living in the public cloud. This leads to added administrative overhead, challenges in troubleshooting end user issues, added bandwidth requirements and costs to list a few challenges.

Licensing challenges with DaaS:

Licensing is one of the key challenges in a DaaS implementation. Microsoft does not offer a Service Provider License Agreement for desktop OS.  As a result, most DaaS implementations are based on shared Windows Server OS or a dedicated Server OS/user. While the shared/dedicated server based DaaS environment could work fine for some users, it does not offer the same level of personalization/customization that true VDI solution based on a Desktop OS offers.

In order for a service provider to host a true windows-desktop-based VDI solution for a customer, customer still needs to own the Desktop OS licenses. Such a configuration leads to increased costs for the customer.

Persona Management:

The degree of user customization offered within a DaaS solution can be significantly limited as compared to on-premises VDI solutions.  These include USB peripheral support, printer management and profile management, to name a few.  Peripherals such as those that rely on a fast USB connection will not be able to communicate quickly with the service provider data center.  Login times can be impacted if some of the profile data lives on-prem.  Printer driver support can become challenging as most DaaS providers rely on universal print drivers.  Print job spooling could consume a lot of bandwidth and impact user performance.  These challenges will be solved over time, but at this point in time, these are significant roadblocks.

Security/Trust:

Most companies are still very reluctant to host their confidential data in the public cloud where they do not have full control.   Certain industries do not allow companies to host data in the cloud due to regulations and compliance laws, because, the question remains – “at the end of the day, who owns the data?”

No Offline Access:

As of today, DaaS providers do not provide any form of offline access.  Users have to be connected at all times in order to access their desktops.  An on-premises deployment (Citrix and other vendors) on the other hand offers solutions that allow the user to access their desktops offline with bidirectional synchronization of data with the datacenter.

I want to summarize by stating that there are definitely use cases for DaaS in every enterprise, but it is not a solution for ALL use cases within an environment and hence not the Silver Bullet.

Follow me on Twitter: @gkuruvilla