What is 'virtualisation sprawl' and how do you stop it?

Gone are the days where virtualisation is just for early-adopters and IT geeks; virtual storage technology has found its place in almost every IT department or data center in the last few years. Whilst there are many advantages in using this technology (fast and almost immediate server deployment, increased server uptime, isolated applications) it also has its disadvantages. One of them is what is known as ‘virtualisation sprawl’. The term refers to the fact that since creating and running a virtual machine (VM) or virtual server is relatively quick and easy; users in companies tend to overstrain their IT department with lots of VMs. At a certain point the number of available VMs and virtual servers is so huge, that the IT administrator cannot manage them effectively anymore.

Users can often create lots of VMs over time and forget about the repercussions afterwards. This leads to lots of unused but available VMs that are taking up valuable storage space. What they usually don’t consider, is that creating a VM (or several) in a virtual server has the same impact as if one were to create a physical server; when creating VMs the same licensing, support, security and compliance issues need to be considered.

Furthermore, having too many VMs poses several technical challenges and risks to the IT infrastructure and its administrators:

1.   Lack of technical control:

• In many cases users are able to create or even delete VMs. One downside is that problems with disassociation of the VMs can occur. This can happen when a VM is deliberately or accidentally removed from the hypervisor inventory.
• When VMs are removed from one cluster to be moved to another host cluster, there is a high risk of them being miscataloged in large storage environments.
• VM fragmentation can occur. When VMs are moved frequently and copying fails, large fragments will be left on the disk or storage.
• Creating huge amounts of virtual snapshots will result in significant storage space use.
• Having not a complete overview over all VMs in the IT environment/company can result in problems when to determine which ones need to be erased. Also in case of data loss it becomes difficult to find distinguish what is production critical and what needs to be recovered.

2. Lack of management – Additional tasks for IT administrators:

• More work for VM administrators: Manage more VMs and juggle with physical resources
• More work for storage administrators as they must provision more disk and storage space

3.   More costs due to unneeded VMs:

• Creating and running VMs is not free! They consume CPU time and memory as well as disk space.
• Licensing costs for the OS, backup agents and databases
• Eventually unnecessary purchases for additional storage space

4.   Reduced performance:

• In some and severe cases too many VMs can also have a crippling effect on the physical server performance. When too many VMs are running side-by-side the overall speed decreases, or when the physical server resources are too low to handle the work load it can fail.
• In some cases data loss can be the unwanted result and since virtualisation creates another layer to the data structure, data recovery is somewhat more difficult in comparison with recovering data from a simple Windows 10/NTFS formatted hard disk drive.

How to prevent virtualisation sprawl

Here are a few ways in which you can prevent virtualisation sprawl:

• Audit VMs. It seems like a simple and obvious proposal, but it’s crucial to understand your infrastructure. Make it a policy that every VM and virtual server must be documented and registered.
• Optimise storage and implement data policies. To prevent the use of unnecessary storage space use technologies like thin provisioning and snapshots. But beware: When you use these technologies and implement them in your data policies, make sure that they are set up correctly. If you run a virtualised server system which is not set up correctly you not only risk losing data or having a system failure, but you also risk an extremely challenging process in terms of getting the data back. In some case data recovery will not be possible at all due to the damage.
• Implement lifecycle management tools. To keep track of all VMs as well as virtual servers it is a good idea to use a lifecycle management tool. With such a tool it is possible to provide VMs only for the users of one specific project or tracking all objects within a centralised database. When a project is finished it then becomes far easier to identify unneeded or unused VMs, which can then be marked for secure erasure and reassignment.
• Implement VM archiving. Many users create a VM just for one project, use it a couple of times and then leave it untouched for months, because they fear that it might be needed again in the future. Instead of keeping these VMs inside the production environment, they should be archived wherever possible. Many backup solutions have the ability to archive unused VMs on cheaper storage systems or tape. 

Erase unneeded virtual machines whenever possible

In addition to these processes, administrators should frequently check if VMs can be deleted and erased for good. Not only because of storage and cost reasons as stated before, but also because of data security reasons. In case of a data breach, having unnecessary data in your production environment increases the risk of restricted or business critical data getting into the hands of criminals. The consequences can be severe: lack of trust by partners and/or customers, huge fines due to non-compliance with data security laws or being victim of criminal activities such as blackmailing. To prevent this from happening, a secure data erasure method should be implemented inside your data security and/or business continuity plan.

Destroying virtual data is different from destroying physical storage; there are several professional tools available which make sure that only the specific VMs are targeted and securely erased while the live environment is still intact and running. Some tools make it easy by allowing you to automate the process and you’ll want to be sure that you’re getting a report for each erased VM, so you can prove that the data has gone. For more information about how to erase different types of virtual data and why it’s important you can check out this article.

Suffered data loss because of virtualisation sprawl?

If you’ve lost data due to the effects of virtual sprawl, it is essential not to panic and to follow the right steps. First, it is absolutely necessary to stop working on the effected VM as well as the physical server hosting the VMs or virtual server. Second, try to terminate the service normally and not by shutting down the server brute force – this could cause further harm to the data. Last but not least, contact a data recovery expert who has experience in recovering data from your system before. Data recovery from virtual systems is not something you should attempt on a DIY basis due to the sheer complexity and risks involved.