Are we there already?? As we hear the many feedback from the clients, it may be… or not. Passed under the radar so, or maybe not appreciated to its true value yet, VMware Mirage, differentiate itself on its own as it defines the way we’ve been looking at disconnected “virtual desktops users”.
- It is NOT a VDI solution!
- But it is part of a VDI portfolio.
- It does not run on a hypervisor
- but it is stored in the datacenter as an image.
- It has datacenter concepts of a virtual desktop
- but is not compute by it.
In a traditional VDI architecture, the virtual desktops are hosted in the datacenter and computed by it. End users leverage a proprietary protocol to access the virtual desktop assigned to them. Nice and clean, nothing on the end point, everything in the datacenter and a communication channel offers the end users a remarkable experience of desktop.
That model is flawless as long as the users is located in a well serviced network where bandwidth and latencies are not in the picture.
This model however shift a little bit when end users are either served by a poor bandwidth or are completely disconnected from the backend datacenter.
Vendors have struggled to find the right balance and have come up with very clever ways of mitigating the experience received by the users. Some have suggested a Type1 hypervisor on then endpoint. You lose the endpoint RING0 operating system, but receive a true virtualized endpoint running images much like vSphere has been doing for the virtual servers. Others have been trying to change the game and minimize the impact such different community of users, the disconnected ones, need and require.
The first attempt was good enough. You were not losing the endpoint OS and were receiving *your* virtual desktop, the one managed in the datacenter, and it was executed by a TYPE2 hypervisor. Clever but efficient.
TYPE 2 & TYPE 1
While efficient, the TYPE 2 model was struggling. Primarily you needed AT LEAST once to be connected to the corporate network and a high throughput network. Think about it: the TYPE2 hypervisor was downloading the entire image on your computer. Some have tried to perform that action over an Internet provider link and found themselves with a very large bill at the end of the month… not a good idea.
So it has a good idea; maybe not a good strategy, surely a tactical response to the competition and market demand.
On the other end, the TYPE 1 model, is efficient but lacks endpoint management. If you wish to leverage such strategy you will lose your endpoint operating system and will have to commit to having a Hypervisor running on the users’ BYOD.
Again there, nice idea, but not sure if users wanted to participated in that disconnected model and comply to the requirements (hardware primarily) dictated. Efficient but not agile.
In essence, MIRAGE, is an end point management solution that empowers IT departments to centrally manage the data while unlocking the image in a true complete executed locally platform. Seems like a long phrase but it’s worth it.
The secret seems to lay into the endpoint management. It Addresses enduser uniqueness, while the endpoint, provided or not by the organization, remains unchanged in its way of functioning and manage the endpoint at a very granular level allowing deltas of blocks to be merged at the operating system, application or enduser profile level. Magic!
No more TYPE1 or 2.
So you’re saying that I can have a computer that I’ve purchased on my own, receive the corporate image while all my settings, applications and personalization remain unchanged?
Yes that’s what I’m saying.
How does it work?
There literally thousands of video and document on the topic and this blog won’t allow me to dog in all of them at once, but in a nutshell, a centralized managed image (OS) is created in the datacenter. The Mirage client is deployed on the endpoint and that same client, “scans” the endpoint and reports it to the datacenter. (http://youtu.be/8oY6OYd0gH0)
Now the magic happen. While keeping the endpoint uniqueness, the backend image is “pushed” to the endpoint in a complementary manner. What is there is there. What is missing will be complemented to comply with the corporate image.
The IT department only becomes the owner of the based image layer (OS) and particularities of that OS (driver library). The rest is either complemented (Application) or remain unique (user personalization)
There’s tons of questions around such model. What about if the endpoint gets a virus, or gets corrupted or data is loss… if you have 42 minutes to educate yourself, take a peak at this youtube video delivered by Cameron Fore http://youtu.be/tuCNYlf61pw
And for the security guys in you, if a virus infects the endpoint, MIRAGE will push back the deltas in the following way: how the image looked like BEFORE the infection, how it looks like now —> what’s the delta? Of course, an AV is mandatory and truly hope that we are aligned on that topic LOL
As far as the applications are concerned… no concerns! Users can safely keep their apps while MIRAGE pushes into the application layer the deltas from the Application Layer CVD Catalog created in the datacenter. Users will end up with their own apps and receive the corporate apps by MIRAGE. Don’t forget, these apps, if required to connect to a backend database, still won’t work ok? But that is another topic.
So? Are we in a VDI model or an endpoint model? Hum… a bit of both I would say. Partially VDI in the way we’ll need to manage endpoints images, and partially endpoint management in the way MIRAGE will manage the integration between the 3 pillars of VDI: user profile, OS and applications.
Now think about the Airwatch acquisition and you’ll get a clear picture of VMware roadmap for the disconnected endusers
Have I captured your attention? Check out http://shop.oreilly.com/product/9781782172352.do
Out of the books and content I have reviewed this book is well done.