How we tested virtual machine environments

13.04.2009
For a product to be included in this round of testing, it had to be able to manage two or more server virtualization platforms from a list consisting of VMware's ESX 3.5, Microsoft's Hyper-V and Citrix's newly 'free' XenServer 5.0.

For virtual machine environments we deployed one Citrix XenServer 5.0 server, two VMware ESX 3.5 servers and two Microsoft Hyper-V servers.

Microsoft SC-VMM was installed on one of the Hyper-V machines.

VMware VirtualCenter and CA's NSM/ASM were running as virtual machines both using Windows 2003 Server R2 on the XenServer 5.0 machine. VirtualCenter was assigned two vCPUs and 2GB RAM while NSM was assigned four vCPUs and 4GB RAM.

DynamicOps VRM was installed inside a virtual machine with 2GB of RAM on the SC-VMM machine using Windows 2003 Server R2.

Insystek's TotalView ran on a separate machine using Windows XP SP2.

The Microsoft Operations Manager 2007 was installed on our Windows 2008 Server domain controller, which was running in a Parallels Desktop 4.0 virtual machine on an Intel XServe hardware platform and was assigned 2GB of RAM. The XServe running MacOS X 10.5.6 also served as our NFS/SMB file hosting machine.

The first VMware ESX 3.5 host and the first Hyper-V host environments were running on an HP DL580G5 server with 16 Intel Xeon MP CPUs (on four total sockets) each running at 2.93GHz with 32GB of RAM. We ran dual-booting processes between ESX and Hyper-V for the tests.

The second ESX 3.5 host environment was running on a Dell PowerEdge 1950 server with eight Intel Xeon CPUs at 1.86 GHz and 4GB of RAM.

The second Hyper-V platform (also installed with SC-VMM) ran on a HP DL160G5 server with four Intel Xeon CPUs at 2.99 GHz with 4GB of RAM. 

The XenServer 5.0 VM platform was running on a Dell PowerEdge 1950 server with eight Intel Xeon CPUs running at 2.493 GHz with 16GB of RAM.

And lastly, TotalView was run on a HPDL140 with four Intel Xeon CPUs at 3.2 GHz with 4.0GB of RAM.

We tapped into the many functions of each product to test its ability to manipulate, control and deliver information about virtual machines in each environment to us in a central location. The type of functions we tested were as follows:

• The ability to create new VMs onto each hypervisor platform.

• The ability to control and manipulate VMs using commands such as start, stop, shutdown, pause, resume, migrate, clone, copy and other commands.

• The ability of to monitor and report information about activity for VM and host.

• The ability to change settings for VMs or hosts.

We then tested how well each VM management tool would assist in securing the hypervisors and VMs by testing policies and user role initiatives. In addition, we tested alerts or alarms (if available) to see when and how information would be distributed.

Besides these features, we also judged the installation and configuration to see how easy or difficult each was to complete.