Towards the end of last year, Microsoft and VMware both made significant announcements around their server virtualisation offerings. Microsoft made the first beta of its Hyper-V hypervisor available for testing, and said the finished product will cost just $28. For its part, virtualisation specialist VMware announced ESX Server 3.5 and version 2.5 of its VirtualCenter management console.

The VMware upgrades deliver some very useful features, such as automated patching, and it will probably take some time before Microsoft can match them. On the other hand, selling Hyper-V at such a low price is bound to attract a lot of interest from firms.

While the obvious attraction of server virtualisation is that it allows the workloads of many physical servers to be run on a handful of host servers, once this is done firms often find that server virtualisation greatly simplifies operational issues. Of course, this translates to lower operational costs and greater flexibility.

For example, the time needed to deploy a new server can be reduced from the weeks it usually takes to buy new hardware down to the few minutes needed to create a new virtual machine (VM). Likewise, virtualisation enables hardware maintenance tasks, such as adding hard disks or rearranging network connections, to be done using simple remote control software.

Virtualisation’s ability to simplify administration was brought home to me recently when I had to add some hard disk space to a VM. The system in question was created three years ago to run my email. At the time I put the mailboxes for various groups of users onto their own hard disk partitions, so if one user’s mailbox filled up it wouldn’t stop everyone else from getting mail. However, all incoming mail is temporarily stored on a different partition until the system figures out where to put it.

Eventually, the partition holding these temporary files filled up with other data, and when that happened, the front-end mail servers couldn’t deliver to the mailbox server. As it was a VM, all I needed to do was shut it down long enough to type one command on the ESX Server command line to expand the virtual hard disk.

Once the disk had been expanded, I could immediately restart the VM. So after only a few minutes downtime the server was running with extra disk space. Updating the partition layout took a little longer, but the entire operation took only about 10 minutes.

Had this server been a physical server fitted with lights out management (LOM) I could possibly have done the same thing without virtualisation. But LOM costs about £200 per server, and with 20 VMs in my datacentre, the LOM cards alone would cost more than the ESX software.