When it comes to new IT investments, the benefits – including cost savings – often depend on how well you use and take care of the technology or infrastructure. Akin to a new car, organisations could easily wind up spending more money if they quickly run the technology purchase into the ground. If your organisation is wondering why it isn’t seeing as great of an ROI on server virtualisation as expected, look to how well the IT staff is managing the physical servers and virtual machines.
1. Use more of your physical servers
Technology and business research firm Forrester reports that one of the chief reasons organisations don’t maximise their virtual infrastructure investment is because they fail to put enough virtual machines on their physical servers. It is tough to strike a balance between too few VMs and too many, as the latter can inevitably lead to poor performance. But setting a strict utilisation percentage is not the answer either, as the organisation might get stuck in a cycle of buying more servers to host the VMs earlier than should be needed.
In Forrester’s research, many companies reported stopping at three to five VMs per server when some of these servers could actually host up to 15 VMs. While how many VMs IT runs will depend on factors such as resource-heavy apps, organisations can typically run three to five virtual machines for each core on a new Intel or AMD processor, according to a CIO report. One Sydney council was able to reduce the number of physical servers in its data centre from 24 to four and save more than $100,000 with a server virtualisation project implemented by Datacom.
2. Avoid virtual sprawl
Of course the danger of creating more VMs on your servers is that you’ll spawn server sprawl. It’s now so easy to create VMs that everyone wants one when they want it. This may not seem as dangerous as physical server sprawl, but it is. Too many VMs lead to an over-allocation of resources, which drives up costs; there’s also the risk the organisation might have to purchase another physical server when they shouldn’t need to. VM sprawl also drains the IT department’s management abilities.
VM sprawl is a sneaky beast – it happens quietly and slowly, so the best defence against it is regularly monitoring how resources are being used and how many VMs are in the data centre one month compared to the next. VMs should have a lifecycle, and careful reporting will help IT departments determine when VMs are no longer needed. Going forward, IT should demand justification for VMs when they are requested to avoid creating them just because it’s easy. IT could also turn off an unused VM every time someone asks for a new one.
3. Replace your physical servers on time
Yes, it can be a drag when you have to potentially fork over thousands or tens of thousands of dollars to replace your physical servers. IDC figures have shown that organisations that upgrade their servers within three-and-a-half-years make back the amount of their investment within 12 months and receive an ROI of more than 150 per cent over three years.Sticking to this refresh cycle can unearth some of the virtualisation benefits your organisation initially sought out, such as improved maintenance and better energy efficiency.
How does your organisation plan to fully realise the cost-saving benefits of server virtualisation?