Everyone is making VDI too hard: it's really just a desktop form factor change

I'm not sure I would dismiss shared non-persistent images in quite so cavalier a fashion.  Especially when you LAYER user profiles and apps dynamically.  Moving from a tightly-coupled array of disparate (and often heterogeneous) unique endpoints (assets) to a loosely coupled, late-bound environment is a pretty ENORMOUS game changer. The same  benefits derived from server virt (independence from specific hardware, and he portability and ease of DR/Backup) is a pretty profound difference in operations. Taking in consideration the effort required to support updates, patches and drivers for unique devices ABSOLUTELY crates a profound change in the way end-user devices are deployed, managed and maintained.  

I would agree that if you build your platform out of all persistent VMs, and forgo any sophisticated Profile Management, then yes, you are really just changing form factors.    In my humble opinion, VDI done PROPERLY demands taking advantage of the miracle of non-persistence and abstraction of the user profile from the physical device.  of course, good desktop management practices still need to be maintained.  However, I would argue that maintaining a master image for hundreds (if not thousands) of end-users is pretty monumental. Toss in a dedupe/cache solution like Atlantis iLiO in the mix, and potentially you are giving end-users a superior workspace experience, particularly   if the mail/file server and database resources have immediate proximity to the hosted desktop as opposed to back ends that must traversed high-latency networks.  Granted, net-new companies can go immediately to SaaS cloud-based apps and spare the trouble of implementing VDI altogether.  Any company, though, with more than a decade of history likely has numerous mission critical, multi-tiered apps which will benefit from a well-architected VDI solution until such time as they can retire / port those apps to cloud.

I agree if you lock down an image, and make use of decoupling via profile management / Application virtualization you have MOST of the benefits of VDI already.  That being said, you'd also have a pretty short HCL and master of SCCM, Altiris or some other such tool to manage and update that environment.  I guess my net-net point is that shared, non-persistent images require a lot less care and feeding than discrete devices, and having the VDI images close to their dependencies  is where true benefits will be derived.  Lacking these, VDI is just moving a problem from one place to another.

We  see the game changer being what Intel has been referring to as Intelligent Desktop Virtualization or IDV.  Because this approach takes things like the ability to control or lock the desktop (referred to as one of the real game changers in the article above), and then adds back benefits for end users as well as additional benefits for IT.

For all the readers who many not be familiar with IDV, here's a little background. Where VDI centralizes both management and execution, IDV centralizes  management but still takes advantage of local execution.  This results in dramatically lower cost, better performance for users, and it gets rid of a lot complexity that crops up when trying to move a workload that was designed to run locally on your desktop into a remote data center.

There are 3 key tenets of IDV that make it a game changer: (1) centralized management with local execution; (2) intelligent layered images; and (3) integrated physical device management (one of the articles on the Brian Madden site rightly pointed out that VDI does nothing to improve the management of the physical devices – IDV addresses this directly).

Server virtualization was a game change for the data center – these days, people rarely ask if they should run a server workload virtualized or not – they just do it.  This is because type-1 hypervisors have gotten so efficient that it just makes sense to do it –  because you get all the management benefits even if you’re not looking to consolidate workloads.  We see IDV doing the same thing on the desktop side – in the future people will just run everything on a type-1 client hypervisor as a part of a complete IDV solution because it dramatically improves the management, reliability, and security of PCs – while still keeping all the benefits like fast, local performance.

Intel has an overview on IDV with more information (www.intel.com/.../intelligent-desktop-virtualization-overview-paper.html%20.html).

One challenge for the IDV category in the past is that it's gone by a variety of inconsistent names and variations.  But now that there's some consistency forming around the category, we'd love to see some articles about IDV.

I can certainly understand Intel's interest in continuing to propagate the use of fat clients with chipsets, RAM, powerful CPUs, etc. capable of running both a Type-1 Hypervisor as well as a guest OS (or two or three.)  At the same time, I can understand that market may desire low-power thin clients with nothing more than the capability to run a browser.  Certainly there are use cases for both technologies in the short term.  As wireless broadband becomes ubiquitous and more apps are moved to the Web, I see less demand for either.

I really hate the term VDI. I don't design 'VDI' solution's I design...well Desktop Infrastructure Solutions... leveraging Virtualization. Ok, so it doesn't really roll off your tongue but my point is ‘VDI’ designs looks something like this.

1) Identify the Use Cases

2) Choose the best method to delivery applications to the users for each use case.

...virtual desktops are in there somewhere

Now if we’re going to go around saying VDI projects are expensive and complicated then let’s make up our mind. If VDI is just a form factor change then actually that’s not so complicated or expensive. What is expensive, is the whole deal...application virtualisation, user virtualisation and tieing it all together. You’d still have to do most of this stuff to layer desktops in the physical world. Take that away and then you’re left with bringing up a hypervisor and some VMs. I sure as heck know I'd rather upgrade 1000 VMs to Windows 8 than replace my physical PCs and roll it out over the network. I do think sometimes we forget the benefits or longer refresh cycles and easy OS upgrades as VDI hasn't been round long enough so see these benefits

So....

VDI is just a form factor changes – agreed if you then agree that VDI is not expensive or complicated.

OR

VDI is the whole layering solution – then sure, it’s pretty expensive and complicated.

Pete

The game changing benefits are about centrally hosting more and building a more efficient compute resource that is accessible from many more places. That enables new use cases, such as M&A, BCP amongst others.

However to do it with current capabilities in anything more that 1-1 is next to impossible or a compromise with RDS, and will result in more cost or lack of flexibility. So the business benefit needs to out weigh the cost.

Better management to enable the desktop to be more like a service is needed all around, but not required to start and that's where I'll bet most people get stuck. For that you need much better capability in the app and user areas, where both Citrix/VMware/Microsoft have next to nothing or a sub par offerings that do not get anywhere close to what is needed the second you reach any level of complexity. AKA the "simplicity" word is being over used with crap solutions like VDI in a box which is nothing more than HDX to a single user desktop/RDS instance and does nothing with the real management aspects. The desktop is complex and distributed so you need to decide if you want to mange this stuff like a service moving forward which is game changing separately from the VDI benefit of centralization. The reality is that many will have to deal with VDI and other models, just like we always have with desktop and laptop. So management across these will get even more complex....