A lot of fingers have been typing about VDI (that's virtual desktop infrastructure) recently. If you're wondering what the fuss is about, the short answer is more than 50 million PCs were shipped last year, which makes it an interesting market for vendors and for organizations that are trying to manage these resources better. Some big players have made noise, such as Citrix (which already has a solution called Presentation Server, err, XenApp), Microsoft (ditto with Terminal Services) and VMware.

The basic concept of VDI is that there are a number of benefits in server virtualization that organizations would like in their desktop infrastructure. To name a few: simpler recovery from hardware failures, centralized management, security... So are there any organizations that are deploying VDI broadly?

In our experience, many organizations are placing desktop infrastructure in the data center for at least a subset of their total environment. We see a lot of Microsoft Terminal Services and Citrix XenApp in our customer base. We're also seeing early evidence of VDI deployments. For smaller user environments, VDI can be as simple as deploying Windows Vista in a virtual machine and connecting using a thin client. We see this a lot in dev/test environments and smaller organizations.

As the environment scales, it's important to think about managability. Brokers can help organizations deploy and manage their virtual desktop infrastructure. I just read a blog over the weekend that describes how to deploy Virtual Iron and Provision Networks to create VDI. And there are other brokers that support Virtual Iron as well, such as 2X Software.

While you can't take your virtual desktop on the plane with you -- at least not yet -- think about VDI the next time your laptop or desktop experiences an issue. You can restore a backup to a virtual machine and be immediately productive. You may find that you don't want to go back.

It’s become abundantly clear that server virtualization is about more than cramming multiple systems onto one server. Consolidation only scratches the surface of what people want to do... there's better business continuity, automated power management, CPU and storage capacity management... the list goes on.

These things may sound easy, but they are actually really hard. For example:

• Making the virtualization easy to deploy - ideally when the hardware powers on it's ready and there's no command line configuration required.
• Integrated storage and network management.
• Support for asynchronous and concurrent operations so that both automated policies and administrators can control the environment without fear of stepping on each other.
• Fault tolerance with event rollback to handle failure events.
• User management through LDAP and Active Directory.
• A rich API that partners can build on.

The list goes on...

Many vendors like to talk about their ability to “manage virtual infrastructure,” but the fact is that other than VMware, Virtual Iron is the only solution that offers these capabilities.

In this blog I’ll use some examples to compare the difference between a complete solution like Virtual Iron versus an incomplete solution like Citrix XenServer. These are basic capabilities that we believe are minimum requirements for true virtual infrastructure management. In future blogs I’ll make similar comparisons to other imposters out there.

1. No Concurrent Operations: XenCenter has no support for object locking and concurrent operations, which means multiple users or policies can modify the same object at the same time. Simultaneous modification in a virtualization environment is as bad as it is in a file-system environment. This can create significant workflow issues and can corrupt the virtual environment.

As an example, imagine two XenCenter users edit the same Windows virtual machine in two XenCenter instances. In one XenCenter instance, the user changes the CPUs from 1 to 2, keeping all other values constant. In the other XenCenter instance the user changes the memory from 512 MB to 523 MB, keeping all other values constant. The user then checks the settings on both XenCenter instances. Whichever dialog was confirmed last overrides the settings of the first without any mention that two users were modifying the same information.

To compare, Virtual Iron locks resources that are being modified as you can see with the lock icons on win 2003 server below, preventing multiple users from conflicting on a shared resource.

2. No Auditability: XenCenter has no user accounts and does not persistently save what changed in the virtualized environment. This means it is impossible to track and record changes to the environment (i.e., who made what changes when). This makes auditing for root cause of failure or compliance impossible.

Note that the log of what was performed is amazingly sparse (i.e., Details: Settings saved). It doesn't tell you what settings were saved, or who performed the action because there are no users saved in the system. This level of reporting is insufficient for most compliance activities.

Second, if you exit XenCenter and re-enter, the information is ‘poof’, gone. Likewise, if you modify settings from another XenCenter client instance, you won't see those modifications logged in the first XenCenter instance. There is no support for multi-user management.

By contrast, in Virtual Iron, the user directory can be tied to Active Directory, OpenLDAP, or eDirectory.

3. No Failure Recovery: XenCenter has no concept of “jobs” (a job is like a database transaction that groups related actions). Without such a framework, there is no ability to roll back actions when a failure occurs.

4. No Automated Policies: XenCenter does not support functional virtualization use cases such as virtual machine restart in the event of failures or dynamic load balancing to satisfy resource requirements. It does not include a policy interface for users or integrators to add custom policies.

Virtual Iron, by contrast, has built-in policies like LiveCapacity that optimizes virtual machine performance and data center utilization by moving virtual machines when resource utilization is above a threshold. And LiveRecovery that ensures high availability of virtual machines without the need for clustering.

On the surface, many virtual infrastructure management checklists look similar. However, it’s important to scratch below the surface to make sure the product has the fundamental capabilities for your environment. Without the basics, virtual infrastructure management cannot support higher level capabilities like business continuity and load balancing that are essential to delivering the true benefits and ROI of server virtualization.