Growing old with virtualisation
If your IT department was quick to use virtualization, you may now have a Virtual Machine (VM) estate that is creaking at the seams, either through old age or capacity problems. Here are some pointers on how you can improve the situation.
The principal question to ask yourself is – “Is my virtualisation estate running efficiently?”
This is best answered by posing two further questions:
- Can you reduce the resources required to run your VMs?
- Is your VM platform up-to-date?
Can you reduce the resources required to run your VMs?
To combat VM sprawl, you should actively look to turn off VMs (or at least reduce their footprint). This can be achieved by identifying VMs that:
- Are no longer being used (and thus could be switched off and decommissioned)
- Do not need the full amount of resources (particularly memory and CPU cores) that they have been allocated and thus could be recovered for other more resource-hungry applications.
A proactive alternative is to ensure that the responsible groups are recharged for the services that they are consuming. This will drive behaviours to promote a better use of the VM estate. Furthermore DMW has seen some success by setting ‘expiry’ dates for non-production VMs and making projects re-apply for access to their VMs.
Is your VM platform up-to-date?
In order to keep your VM platform up to date, you could consider:
- Hardware refresh of hosts (there are tremendous step changes in performance that can be achieved when comparing today’s Intel/AMD hardware with that of five years ago; for example virtualisation ratios have changed from 1:5 to up to 1:20 since 1997)
- Hypervisor version refresh (or at least patching of the current version)
- Newer technologies (thin storage provisioning, better network tagging and management of VM traffic)
In summary – providing a VM-hosting infrastructure is still a very good choice for most people. It continues to offer significant cost-savings opportunities for both development and test as well as production environments. The steps taken above should, of course, be feeding into a larger ongoing capacity management activity so your infrastructure continues to serve you well into the future.