Taking Virtual Desktop Optimization Too Far

I’ve been spending a lot of time focusing on the lessons learned with Windows 7 migration. There are plenty of articles about optimizing the Windows 7 image (I’ve even authored a few of them in my Windows 7 Optimization section). However, I am left to wonder if some of these recommendations go too far.

Most users want to personalize their desktop. They want to configure their own backgrounds. They want to configure their themes. They want desktop sounds to alert them of email messages. So why are we recommending that we disable all of these features?

To save resources. Yes. That is the answer, but is it a good strategy? Disabling this functionality will save some network and system resources but the cost is user acceptance. User acceptance is going to make or break the desktop virtualization solution. You don’t believe me? Look at your own desktop. Did you customize the background? Sounds? Themes? What else? What would you say if you couldn’t do these things?

What would you say if Microsoft or Apple lacked this personalization functionality? They wouldn’t hear the end of it. They would be criticized by every person that touches the OS. So why am I seeing so many Windows 7 optimization guides recommend doing just that?

There is nothing wrong with optimizing the desktop, but these optimizations should not impact the overall acceptance level from our users.

Daniel – Lead Architect

6 thoughts on “Taking Virtual Desktop Optimization Too Far”

  1. Daniel,
    I agree but look at it from a little different perspective. Of course I’m no expert but looking at Ron Oglesby’s latest post on Brian Madden. http://www.brianmadden.com/blogs/ronoglesby/archive/2010/09/22/does-os-quot-tuning-quot-help-vdi-performance-part-1.aspx It seems running all of the optimizations reduces load by 9%. While it could be argued that 9% is a significant number is it worth giving up the user experience? More importantly with the new hardware available such as the Cisco B231 with 248GB of RAM and so many cores I lose track 9% is just not worth it. In the end Moores law compensates for the 9% every couple of months anyway.


  2. Good point regarding Moore’s law and the minimal impact.

    Maybe I need to create Feller’s Law: The probability of user issues increases in relation to the number of infrastructure tweaks. 🙂


  3. This has to be a cultural thing that differs from one organization to the next. At a company that has more relaxed policies users may expect to treat the desktop environment as their own, free to customize, tinker, and tweak at will. In a more structured corporate environment where everything is plain vanilla and locked down tight, users won’t be surpised if their new Win7 VDI desktop is both of those things.

    We’re actually looking forward to VDI as an excuse to corrent some policy mistakes from the past and start to rein in some of our free spirits who’ve been given a bit too much freedom over the years (often for political reasons). The notion of “common” and “shared” helps us deal with pushback from people who’ve staked a claim on the desktop.


  4. @johnm

    Might I ask why, if you want a locked down environment, would you select the VDI model instead of the hosted shared desktop? Usually that is one of the deciding factors I’ve seen for going with hosted shared desktops.


    1. Semantics — I was using “VDI” in the generic sense, where a user’s desktop environment is virtualized on a shared hardware infrastructure, not specifying a particular choice from a vague list of classifications.


  5. I still do not understand why some business would someone want VDI model 🙂 (So this should be clarified first.)

    What are the usecases?

    1. The standard usecase implicates following conditions (correct me if I am wrong):
    A. Centralized/Serverbased (extra costs) AND
    B. Statefull (extra costs) AND
    C. Full admin/unmanaged (=arbitrary installations and system modifications) <– Do you really need this for people who work in the way they do not need laptop?
    (Colleges, maybe?)

    Risks: Should any of A, B or C fail you automatically fall-back to:
    I. XenApp (It's highly managed by default and can offer stuff like visual customizations, sounds etc. as well. Honestly I think that it offers un precedented user-density per GB ram and per CPU, etc.) OR
    II. Normal desktop (Can be managed or unmanaged.)

    2. XenClient + Synchronizer (+ XD.)
    Okay, using these in conjunction might be nice.

    3. Service providers:
    As a customer: Order 50 desktops, get 50 desktops.
    As a customer buying DaaS you can decide to manage it in your own way, using your own tools. Infrastructure itself is ran by sb. else who doesn't (need) to care about OS(/application) management.
    But again: How often does it happen that you need to buy DaaS and you do not need to buy end devices?…

    Well, in short I think that VDI-based products will still have to prove their usefulness a little bit more. At this stage it rather looks like a tactical solution for some marginal cases… (With not so sure business case…)


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.