Today over at the Gartner.com blogs, leading analyst and all-around top bloke Gunnar Berger penned a great post on the announcement last week by Amazon (at re:Invent) of their Amazon Workspaces offering. Put perhaps too simply, this new solution uses the famous Amazon Web Services (AWS) backend to provide a so-called ‘Desktop As a Service’ (DaaS) solution, allowing end users to connect on-demand to their own desktop in the cloud, hosted on Amazon’s servers.
The post – ‘The DaaS floodgates are open thanks to Amazon WorkSpaces‘ – is well worth reading, and one of the best balanced views I have seen so far. I am so happy to see an even analysis that attempts to get to the roots of this offering with real world insight. Seeing so many boosters posting uncritical reviews from re:Invent was driving me crazy, so it is great to see a more balanced view of Workspaces.
I am particularly pleased that someone of credibility is finally pointing out, for example:
Technically speaking there was absolutely nothing new in this announcement. Many DaaS service providers have been offering near identical services for years.
As Gunnar notes, there are already multiple, mature, enterprise-grade offerings available from great enterprise vendors like Citrix and VMware (via Desktone – which actually owns the trademark on the term ‘Desktop As a Service’), and many others. In my opinion, especially for core VDI use cases in larger enterprises (e.g. healthcare, DR, process work (CSRs, claims, etc)), such existing offerings are not just alternatives, but probably much more attractive than Workspaces. More on that below.
Moreover, I agree wholeheartedly with, and am thankful for, Gunnar’s spot-on opinion that …
Anyone in the VDI space knows that VDI cost too much, the licensing is broken, and it’s too complex.
I do, however, take issue with the statement:
DaaS fixes everything but the license issues.
Again, I appreciate Gunnar’s insight on the inclusion of the Teradici tie-in. That is surely a smart move that will address some of the more egregious problem with cloud-hosted VDI. However, in my opinion PCoIP is perhaps the second-best VDI protocol available, and it will not solve even all the most pressing usability problems of DaaS, let alone all its problems in general.
Even Citrix HDX – which I consider to be more mature protocol with better capabilities not just for HD media streaming and screen compression, but also for redirection, multiple client support, input handling, touch support, and even distributed processing for operations like GPU acceleration – does not solve all the problems (bar licensing) of such an offering. Why would a business deliberately opt for a technology that doesn’t even go that far?
Then there are many other issues that have held back enterprise adoption of VDI/DaaS, and public commodity cloud in general. The offer of (let alone the ability to negotiate) security and compliance assurances, Service Level Agreements, performance guarantees, response times (human as much or more than system) – not to mention a legitimate way to calculate downtime that does not depend on the client’s application architecture – make existing offerings more competitive, not less. Amazon is notoriously inflexible on all of these factors, which are mission-critical for enterprise IT.
The lack of customer support will be another major barrier to large scale adoption of Workspaces. It may be acceptable to an enterprise data center customer to fend for themselves in the face of technical issues – and DaaS will have even more of these than even VDI, which in turn has more than a legacy hardwired desktop. However, this ‘fanatical lack of support’ will not fly at all when end users are brought into the mix. Even less so for those small and medium business that should be the core customers for Workspaces, but which have no dedicated IT support of their own. And if Amazon will not provide end-user on-call support for these cloud-based desktops (which they almost certainly will not), how does it expect enterprise IT organizations to support them?
This actually highlights an even more fundamental issue, that perhaps warrants even deeper consideration – desktop virtualization is inherently much more complex and more difficult than server virtualization. Just because a business (IT department or service provider) is an expert at one, is no guarantee it can be successful in the other. Indeed, it may ensure exactly the opposite, as bringing server assumptions to a desktop deployment is like bringing a knife to a gunfight.
This not coincidentally leads into another issue with Workspaces – the use of a server OS to deliver a desktop environment, seemingly only to ‘get around’ the licensing issues (which I only learned of through Gunnar’s post). Gunnar is on the mark when he writes:
I don’t think Microsoft ever intended for Windows Server OS to be used in this way …
But again, I take issue when he goes on to conclude this statement by saying:
… but hey it works and it saves money.
From a usability perspective alone, this is precarious, especially for those core ‘process worker’ use cases for VDI/DaaS. Training low-skilled process workers on using a desktop delivery of a server OS will be counterintuitive at best. There are likely to be application compatibility issues between desktop and servers OSes too, not to mention the potential for issues with client-side devices and driver support. And how will businesses adopting Workspaces be able to harden all the server-side services that come with a Windows Server OS that, in the desktop use case, are at best an inconvenient use of resources, and at worst a massive security hole? Or, if Amazon does that for them, how will they know and trust Amazon’s configuration when it will (undoubtedly) never be revealed to them?
It also seems to me that DaaS in any case solves a short-term problem that not that many people have today, and fewer will have tomorrow. Startups in particular are much more interested in BYOD solutions than VDI solutions. They would rather provide no desktop to their staff at all than provide a desktop in the cloud (not to mention their devices are much more likely to be running OS-X, not Windows – a key limitation of the Workspaces offering). Meanwhile VDI/DaaS has not excited most enterprises either (notwithstanding a handful of large-scale adopters that are the exception, not the rule). Like the rest of AWS, Workspaces does nothing to change that.
Plus, for any new adopter, better options are already available in the guise of what I once called application virtualization (notably Citrix XenApp), not to mention straight-up cloud-based applications. In very short order I expect such alternatives to provide better (and ‘cloud-native‘) solutions that are going to be more usable and more future-proof than either VDI or DaaS, and which will leverage the growing instance of personal and consumer-styled devices and BYOD approaches.
In these respects and more, therefore, I see Workspaces in particular, and DaaS in general, as close cousins to the old ‘Internet Kiosks’ that today sit 99% abandoned at airports throughout the world. They were at best a patchwork solution, delivered too late, with too many fundamental flaws, unable to keep pace with technology, and left behind by the consumerization of IT. As a result, they were ultimately and inevitably doomed to a limited lifespan at best.
Fast forward to today, and no-one uses Internet Kiosks because smartphone evolution has mean that almost everyone has superior interfaces, applications, and usability with mobile broadband in the palm of their hand. Similarly, as consumers increasingly drive business technologies, and their businesses increasingly embrace BYOD and consumer-driven IT, there is simply not going to be enough of an ongoing need for Workspaces to thrive in the medium- to long-term.
Now, I have already learned a lot from Gunnar’s blog, so I will not be surprised if he schools me on any ignorance or misconceptions that I may have shown here. After all, I have been substantially removed from the desktop and application virtualization space for a while, so I might be off-base on some (all?) of my opinions. Moreover, I would not pretend to think I could match Gunnar’s knowledge of this space regardless, considering he lives and breathes it every day, and I am at best a casual observer.
Plus in any case, this is early days for Workspaces, so anything could happen. I am clearly skeptical, yet still interested to see how it plays out. After all, perhaps ‘the Amazon halo’ can in fact make this work, in the same way that Apple made portable MP3 players work despite the iPod’s obvious flaws, and the raft of technologically superior extant competitors.
Finally, I think many would agree that, at very worst, Gunnar definitely has a fighting chance to be the coolest guy I’ll ever know.
So, at least we can agree on that! 
