I am going to apologize in advance for the length of this post. It represents several years of discussions and ruminations that have been percolating in the brain…
Let me be clear – I am a fan of VDI, in particular VMware View (which, by the way, VMware continues to improve and add well thought out features to…. View 4.5 looks great, and they will have an iPad client soon. They previewed it at VMworld this month?). In my personal opinion, centralized desktops make a TON of sense for corporations looking to make their operations more efficient, decrease their support costs, and increase access and availability. In addition, the supporting technology provides a nerd like myself lots of tasty morsels to sink my teeth into like server virtualization, application virtualization, firewall traversal, SSL certificates, and so on. It’s a meaty subject, and I can really nerd out on it. So what’s the problem?
I spent several years of my career focused on Citrix MetaFrame (later Presentation Server, now XenApp… Another awesome technology, though its use case is somewhat more limited). During my tenure with Citrix, I discovered that a centralized desktop management model requires a unique blend of skills to implement and maintain, including server config and maintenance, desktop management, application installation and troubleshooting. This means an investment in the team maintaining the solution to not only get trained in the product, but a couple other ancillary disciplines as well. More problematic, such a solution requires a significantly greater amount of effort just to get it in the door… And all this is the same for virtualized desktop solutions! You have to coordinate efforts with, and get buy-in from more teams:
- the server team
- yet another set of systems for them to set up and maintain
- the networking team
- there aren’t a lot of server network segments that are set up for DHCP and dynamic DNS in the core of the Datacenter, both of which are required for most virtualized desktop solutions
- the security team
- you will be allowing users to cross several security perimeters to get access to possibly hundreds or thousands of desktops which will now be in the Datacenter, as opposed to a physically separated environment
- the destkop group
- they will have to figure out new ways to service the desktop, and until everything is rolling, and probably for a long while after that, they will have to split their operations between the physical PCs and the virtual PCs. AND there’s a good chance they may never get to a 100% virtualized desktop environment.
- the users
- Well, you may not actually have to convince the users, but usually there is at least an individual, if not a team, focused on the user experience. They will frequently take the stance that the user experience “has to be exactly the same” …. It won’t be, believe me. It may be better, it may be worse, it WON’T be the same.
- the financial team
- This could be the CIO or someone else responsible for the yearly desktop budget. They are usually looking for an immediate reduction in the cost of deploying desktops by buying thin clients. Well, thin clients aren’t necessarily all that inexpensive, and don’t forget the servers, network, and storage that will be needed for the project. Oh, yeah, and the software and the training… And the time to figure out and integrate new operational procedures across all the teams I just mentioned… Doesn’t sound like a money saver, does it?
In Part 2, I set forth some thoughts for how to approach your desktop virtualization project with a clear head and free of misguided expectations.
In Part 3, I try to help you, the reader, understand some of the traps that are easy to fall into when evaluating your solutions.
– Posted using BlogPress…