Are Zombie Virtual Machines tying up petabytes of your storage?

Virtual Machine (VM) administrators still need to be extra vigilant against VM sprawl as more and more users “self-provision” to set up their own virtual machines. VMs are software implementations of computers that execute programmes like physical machines, and they are the enabling technology behind cloud computing.

The ease with which VMs are created and the lack of discipline in removing them when no longer needed means VMs are needlessly using up valuable resources and generally creating ongoing performance problems.

With cloud computing, end users can set up VMs on their own without the participation of IT management. The scale of this problem becomes apparent when you take into account VMware’s news in July that there are over 20 million VMs running on vSphere alone, and someone turns on a new VM every six seconds.

Even conservative calculations show petabytes (1 + 15 zeros) of storage are being wasted, and that costs tens of millions of pounds. Virtual machines are very cheap and easy to set up, and many are created for temporary use and not because there is an ongoing need for the business.

Unfortunately these “zombie” VMs are often left to run on and on when not serving any useful purpose, and that wastes valuable resources.

While most IT managers believe they are too busy handling maintenance, upgrades and troubleshooting to police unused virtual machines, Hipperson suggests that administrators could use automated tools to undertake systematic audits of all VMs and identify those that haven’t been used recently – a month or longer, for example. Then the owner can be queried to determine if the VM really needs to be kept.

Of course, there are crucial differences between server “utilisation” and actual “usage” that need to be differentiated. This would be a risky strategy if important data and documents were threatened with deletion, even if they haven’t been accessed recently, so a review policy will focus on truly wasted space.

In addition, hierarchical storage management can be used, so important, but infrequently accessed information, can be moved to lower cost storage devices. Tiered storage will free up the most expensive storage resources, and when combined with eliminating truly unnecessary Zombie VMs, enterprises can improve their return on investment from virtualisation.

Stace Hipperson, Chief Technical Officer of Real-Status, has over 20 years’ experience in IT networking, process, sales, project management and support. At Real-Status, he is responsible for all aspects of the technology vision and IT architecture. Stace initiated the visualisation product suite, which is led by HyperGlance, and he is instrumental in guiding the product strategy for the products.