i take a half holistic google +Bradley Horowitz
and salesforce +Marc Benioff
approach to application delivery and platform being wholly cross platform through the browser
the other half of my approach is one learned from years of field service/support engineering various degrees of most disciplines few who haven't worked in small shops and failed startups ever need to be such generalists. when you are lucky to even have a client that wants something done badly enough to pay for it, you get such a mismash of working on crappy home-office computers and gov ed/muni/big cannon enough to shoot a gnat situations... you learn a lot of tricks most never have to. that small tech consultancy hustle is normally bottom of the barrel burnout material - or people who struck out in the .com boom and have too many obligations to take risks again.
i got in this situation because i am crazy enough that i never could hold down a job for long (record is two years but i normally managed to stay in the saddle. had to freelance and create my own opportunities because my reputation was shaky enough skills had to speak for themselves. i normally got jobs by being on-site or remote with a client who was impressed enough to give me a better offer and think i was a steal at that price... until a few months or a year later when they got tired of the fact i was so autistic and bipolar that i often was psychotic even when not drunk and high which was infrequent.
sorry for the life story biography crap, but the fact i was so unstable that i fell in and out of more disciplines yearly than most people change jobs has given me much more of a 'big picture' insight and thousands of failures to choose from. i got fired, screwed up deals, bungled projects or sometimes even entire programs so bad... that i always managed to get myself over my head having to talk my way into a gig where i sunk or swam, but having to live under the constant pressure of proving i was worth tolerating being nuts enough to be certifiable - led to prodigious talent when enough years passed that i found myself being a scotty style miracle worker.
the fact i wore nearly every hat that one can and normally had to rely upon ingenuity than money/reputation because only folks that would hire me were small enough to risk taking me at face value gave me a wide array of telecom/carrier/enterprise/smb consulting most people never learn because they are sane/stable/dont change jobs like most change underwear. jack of all trades and master of none is the saying, but normally better than those who know just one.
so i received the benefit of 25% big iron/telecom/major responsibility with the remaining 25% inredibly painful lessons in startups and tiny businesses without any capital. duct tape and zip tie engineering @ 80 miles an hour for clients that wouldn't spend money to do it right and threw more good money after bad than just doing it right the first time is why all this 'cloud' stuff makes so much sense to me.
so sorry to go on a long tl;dr tangent but that's kind of my trademark move, and hope that it gives enough context that the simple answer to your question makes sense. why the hell would you pick getting locked in with citrix or vmware for application delivery in a proprietary manner to have to pay licensing simply to have less of a feature set and scalability than the tried-and-true client-server web/app/data tiered approach? premise cloud is just as simple of a concept for a 2k8r2 instance with all your software/uptime/redundancy/scalability safely in rackspace's hands (which dollar for dollar beats having them in your own be you one man show or fortune 500) that a single engineer can manage entirely on their own with cloud load balancers and spinning up instances from a known good state? http://techcrunch.com/2012/04/12/how-to-scale-a-1-billion-startup-a-guide-from-instagram-co-founder-mike-krieger/
look at slide 177:
2 backend engineers can scale a system to 30+ million users.
billion dollars. 30 million users. 2 engineers managing web/app/data tier. what this entire industry has been about for the longest time was making massive scale work, and if you have any in-the-trenches consulting experience you know most clients do it horribly be them microsoft or joe's fishing tackle supply store. it used to be the amount of iron/engineers/pipe/infrastructure you needed to pull off terminal services for an organization was too much capex to even consider with the roi. when your opex is based upon purely utilization and you don't have to cough up the licensing or cut that huge po and hire that staff, all of a sudden you can create a formula that always creates a neatly balanced equation.
a few years ago it would make sense to run with this vendor, but i bet even with the cals and exchange doing it on rscloud and creating the necessary infrastructure to make it fully client/server and scale from 1-millions of users would take less time and cost less money.
didn't even read the blurb about this solution or how it works, simply know that it's as outdated as writing a check at the grocery store or putting a tape in your stereo to hear your favorite song. :)