The concept of a dynamic data center relies on the ability to increase and decrease service-oriented workloads based on actual business needs. When the business needs peak, you must be able to dynamically increase the workloads you run to meet demand. When business needs are at a low, you should also be able to decrease the number of service-oriented workloads you run to minimize ongoing costs.
While IT manufacturers have been touting the arrival of the dynamic data center for some time, organizations have not been able to move to this model until the advent of one single technology: server virtualization. In an x86 world, server virtualization gives you the ability to run as many workloads as you require and to provision or de-provision workloads as needed based on actual business need.
Take the following example: You run a payroll application (see Figure 1). This application is "parked" most of the time. Human resources personnel may need to modify records and obtain data from the application, but this level of workload is very minimal in terms of the resources it requires. However, twice a month you need to actually run the payroll and generate employee payments. When this batch process runs, the payroll application requires more resources -- many more resources.
In a virtual world, you can "park" the application on a host server during the time it requires few resources. Then, when it needs more resources to run payroll, you can dynamically move it to another host server and make sure the application has the resources it requires for this job. Finally, when the job is complete, you can move the app back to a parked state.
For this process to work, your hypervisor infrastructure must be able to perform two tasks:
- Move virtual machines (VMs) from host to host without service interruption.
- Support either the assignation of a variable amount of resources to a VM, or modify the resources assigned to a VM during its operation.
While Microsoft Hyper-V does not yet support the ability to move VMs from one host to another while they are running -- the ability will be a part of Windows Server 2008 R2 -- it does have the ability to modify some resources on the VM while it runs. For example, you can modify the weight one virtual machine will have against others while accessing the host server processors during operation (see Figure 2). Modifying these values would change the way the Hyper-V host assigns resources to this and other virtual machines.
Other hypervisors have more capabilities than Hyper-V in terms of variable resource assignation, but the fact that Hyper-V has some of these capabilities in its very first iteration shows that Microsoft is serious about creating a hypervisor that fits into the dynamic data center. Future iterations of this hypervisor will see the addition of features such as dynamic memory management within virtual machines, hot-adding of resources to VMs and more.
Is Microsoft Hyper-V ready for the dynamic data center? The answer is no, not yet -- but it's on its way. There is no reason for your organization not to implement Hyper-V today as part of your move to the dynamic data center. You won't be able to support the payroll example just yet, but you will in time. In the meantime, you can benefit from the physical server consolidation any hypervisor offers.
THE BIG QUESTIONS SURROUNDING HYPER-V
Can Microsoft really make an impact with Hyper-V?
What does it have to offer?
How does Hyper-V rate?
Can it meet high availability requirements?
How does it fit in the dynamic data center?
The bottom line
ABOUT THE AUTHORS
Danielle Ruest and Nelson Ruest are IT professionals specializing in systems administration, migration planning, software management and architecture design. Danielle is Microsoft MVP in Virtualization and Nelson is Microsoft MVP in Windows Server. They are authors of multiple books, including the free Definitive Guide to Vista Migration for Realtime Publishers and Windows Server 2008: The Complete Reference for McGraw-Hill Osborne. For more tips, write to them at firstname.lastname@example.org.