The Long View: Desktop Management

With a billion virtualization and cloud solutions in the marketplace right now, developing a cogent long-term desktop management strategy is proving to be a challenge.  We have a lot of specialized equipment in the research areas of the organization, including a few Windows 98 computers that we can't decommission; a hundred task-oriented workers that could easily move to Terminal Services; a couple hundred "power users" that need some horsepower at the desktop for digital imaging and radiography; a couple hundred clinic machines whose loadsets and roles in the practice are still developing rapidly.  With this kind of diversity in the enterprise, having a simple and sustainable support and management model isn't easy.

If it weren't for the recent push to reduce power consumption and costs as much as possible, it would be tempting to expand our single-loadset, group policy-based model to the entire enterprise.  This would maximize the amount of cycles at the user's desktop, but also maximizes  the amount of support necessary for all that hardware.

On the other hand, moving all the desktops to the datacenter using Terminal Services (I guess it's renamed Remote Desktop Services now) isn't practical because of all of our faculty, research, and clinical special needs, so we're stuck with a hybrid solution; necessitating two support models and extra complexity.  In general, the sysadmins manage and troubleshoot the thin clients and TS farm, and the desktop support guys manage the fat clients.  

Michael Greene (Microsoft higher-ed and virtualization guy) confirmed my hunch during a recent conference call (that I'd scheduled hoping that he would have an obvious solution at the ready, which was a ridiculous expectation in hindsight) - the hype is certainly around moving the desktop cycles to the datacenter and the disk to the cloud, but the technology isn't fully baked yet, and besides, it's not for everyone.  Additionally, the super-new hype is around the smartphone/mobile device replacing the desktop machine relatively soon.

After all this, I don't feel like we've made much progress toward a workable 5-year plan in this space.  Therefore our current solution looks like:

• Upgrade the TS farm to 2008; maybe add as many users as we can and decommission their desktops
  
• Keep imaging and replacing special purpose and power-user desktops according to replacement schedule
  
• Refine the GP-based UX provisioning that we have in place
  
• Continue to use a single WIM loadset across all of our fat clients
  
• Continue to waste lots of power and desktop support man-hours

To paraphrase every infomercial ever, "there's got to be a better way!"