Virtual disk performance and VM snapshot issues in Microsoft Hyper-V

As he dips his toes in the Hyper-V waters, VMware expert Mike Laverick finds some downsides to the VM snapshot process and virtual disk performance.

This article is part of our series chronicling VMware expert Mike Laverick's experience with Microsoft Hyper-V. He took a Microsoft virtualization course to broaden his horizons, and in this part on what he didn't like about Hyper-V, he talks about problems with the virtual machine (VM) snapshot process and virtual disk performance.

Like VMware ESX, Microsoft Hyper-V uses what are called snapshots (if you use Hyper-V without System Center Virtual Machine Manager (SCVMM)) or checkpoints (if you use Hyper-V with SCVMM).

On the surface, the VMware and Hyper-V VM snapshot implementations are very similar. Each VM snapshot process captures the memory state of the virtual machine (VM) at a specific point in time, and they create a delta or differencing disk to keep track of changes made after the snapshot is taken.

More on virtual disk performance
Choosing a virtual disk type for VMware ESX

Optimizing virtual hard disk performance in Hyper-V
However, I discovered in the course a major downside to the Microsoft implementation: To delete a VM snapshot and merge the contents of the delta disk into the base virtual disk, the VM must be turned off. In Hyper-V, the VM snapshot can be removed whilst the VM is powered on, but it merely removes references to the snapshot in the user interface. The underlying technology is still engaged, and the delta file continues to grow.

The VM snapshot can be there, but you simply cannot see it, and a number of my fellow students reported that it caught them off guard. It isn't until the VM is turned off that the snapshots are properly removed.

In fairness, many VMware customers have been caught off guard by not monitoring their snapshots. It was only with vSphere4 that VMware customers received alarms and alerts about the growth of their snapshots. And many still run PowerCLI scripts to generate daily reports about their VM snapshot usage.

Thin virtual disk performance
Both ESX 4 and Hyper-V R2 have the ability to use thick/fixed-size disks as well as thin virtual disks. With a thinly provisioned virtual disk, the VM's hard disk size might appear to be 40GB, but its actual on-disk size will only reflect the volume of data -- say, 10GB.

Thin virtual disk provisioning allows for phenomenal savings on disk space. VMware has long made concerted efforts to demonstrate that its thin virtual disk performance is no different than its fixed-size disk performance, but that hasn't been the case with Microsoft until recently.

Microsoft admitted that Hyper-V R1 thin virtual disk performance suffered and increased fragmentation. But with the release of Hyper-V R2, Microsoft is now claiming near-native virtual disk performance.

Regardless, the sad reality is you may still be forced to use VMware's Raw Device Mapping or Microsoft's pass-through disks to meet the performance requirements of third-party ISVs, such as Oracle.

About the expert
Mike Laverick (VCP) has been involved with the VMware community since 2003. Laverick is a VMware forum moderator and member of the London VMware User Group Steering Committee. Laverick is the owner and author of the virtualization website and blog RTFM Education, where he publishes free guides and utilities aimed at VMware ESX/VirtualCenter users, and has recently joined SearchVMware.com as an Editor at Large. In 2009, Laverick received the VMware vExpert award and helped found the Irish and Scottish VMware user groups. Laverick has had books published on VMware Virtual Infrastructure 3, VMware vSphere4 and VMware Site Recovery Manager.

Dig Deeper on Microsoft Hyper-V and Virtual Server