Application virtualization, virtual desktop infrastructure and all topics relating to centralizing desktops via virtualization were hot topics at the Server Blade Summit in Anaheim, Calif., last week. SearchServerVirtualization.com met with Neil there to discuss the topic.
SearchServerVirtualization.com: Did the interest in centralizing desktops via virtualization catch Microsoft by surprise?
Mike Neil: I'm somewhat surprised at how quickly companies are investing in these technologies without the proven track record. It may be a little bit of the tail wagging the dog scenario, where the company is making the decision to centralize due to various problems and then thinking about virtualization for solving those problems, rather than the other way around.
We were fairly nimble in reacting with the
Will the option of virtual desktops slow Vista migrations by businesses?
Neil: Business users are typically cautious, and they're looking at the technology to see if it meets their needs, particularly if the application compatibility is there.
The new scenario, I think, is going to help us. If you're going to go about deploying an architecture that is significantly different than what you've done in the past, you might as well be trying Vista as part of that solution. It's sensible to migrate to something you probably were going to use anyway.
The other piece that makes this migration much more viable is the SoftGrid technology, allowing for that application isolation and not having to deal with as many app compatibility and interaction problems. Being able to stream apps to those virtual machines works very well.
The combination of moving to Vista and then SoftGrid for application deployment makes a lot of sense if you're going to be making a radical change in the way you architect your client systems, moving them into a centralized environment.
I've heard analysts and IT managers say that virtual desktops will make it easier to migrate to Linux? What's right and wrong about this assumption?
Neil: The customer is going to choose the platform that has the applications and technologies that they need to do their business. The application base and capabilities of the Windows platform has made it dominant in the client area. I don't expect that to change.
So, you expect the client operating system market share to stay the same?
What virtual desktop technologies do you think are most viable now?
Neil: We see a couple of different use scenarios emerging right now, based on Terminal Server and virtual machines (VMs).
Obviously, today Terminal Server is a widely used for centralizing applications in the data center and remotely presenting and accessing them, and we've done a lot of work in Longhorn Server to enhance capabilities like access from outside the firewall. Terminal Server via our partnership with Citrix is behind most of the centralized desktop deployments today.
The emerging technology that's interesting is using virtual machines with centralized enterprise desktop licensing to enable that. We see three scenarios in the virtual machine side emerging:
- You, the user…get(s) a virtual machine on a specific server;
- Your desktop connects to the connection broker which spins up a virtual machine within the pool of physical servers that allow you to connect to applications. That environment can be more dynamic.
- More on the edge is a model where you connect to the connection broker, and it creates a [virtual machine] from scratch for you. Then, you use technology like SoftGrid to stream applications down into that environment.
All of those right now I would characterize as being cutting edge scenarios.
Will the additional planning needed to implement and manage centralized, server-based virtual desktops pose more challenges than the status quo, the rich-client model?
Neil: In some ways, yes. You'll need to be much more cognizant of the work habits of all your users. You could centralize and standardize and end up reducing the flexibility and productivity of your end customer and potentially impact the business negatively.
You'll have to spend the time to understand this model. Moving into the virtual desktop environment brings substantial upfront investment costs. There's pain in migrating your user base; training that needs to occur.
Many people have and will underestimate the network requirements, both on bandwidth and availability. If you're in a geographically-dispersed environment, you have to make sure that you have high-speed WAN connectivity between the environments and that they're highly reliable. Obviously, if you can't get to your centralized desktop, you can't get your work done.
So, it seems that – other than backing the Terminal Server-Citrix model – you're not advocating a quick move to server-based virtual desktops?
Neil: The centralized desktop is an interesting area customers should explore, but I wouldn't rush into it at this time. It's not as well proven as it should be for mass adoption. The early adopters will provide a lot of data for us, but I don't think the mainstream should rush into it.