A virtual environment seemed like just what your organization needed, didn’t it? Almost unlimited resources, easy spin-up/spin-down of virtual servers, and big savings on hardware. The experts said so, your colleagues said so and even the analysts said so. You did all your homework and started sailing on that virtual sea.
So what happened to your ROI? Why isn’t your virtualization strategy panning out the way you expected? Why are you seeing spikes in both operating expenses (OpEx) and capital expenses (CapEx)?
After hundreds of conversations with customers, we’ve concluded that the answer to those questions lies in one simple fact: It’s so easy to create and distribute VMs that many companies overlook the need to manage them.
In this series of three posts, we’ll explain some of the concepts you need as you optimize virtualization management in your organization.
You can create and run dozens of VMs per physical server. Just keep in mind that even though the workloads are virtual, the server and storage resources are not.
VM density is an important metric because you trade it off against optimal performance. If your density is too high, then your VMs are competing for precious resources, and the most common symptom is poor performance.
That’s a problem, but it’s easier to find than low density. If your density is too low, then you have physical resources going underutilized. The most common symptom is excellent performance, which makes it look as though there’s no problem at all. But once you edge back toward high density, you’ll see performance problems. Striking the balance is what optimizing virtualization management is all about.
VM size is also a factor in density. Creating immense VMs leads to inefficient sharing of disk space, physical memory and CPU.
Physical sprawl is pretty easy to spot: rows of servers take up a lot of real estate and cost a lot of money. VM sprawl is less conspicuous and can be more difficult to identify. Almost every virtual data center exhibits some symptoms of VM sprawl:
- Abandoned VM images — They’re no longer in the inventory, but they’re still in your virtualization environment. Even though they’re invisible, they consume resources.
- Powered-off VMs — Nobody has started them up for months (years?), yet they still take up physical disk space.
- Unused template images — If your organization has published templates for creating VMs, are you sure anyone uses them anymore?
- Snapshots — VM snapshots that go a long time without modification are also disk hogs. If they’ve outlived their usefulness, then they’re candidates for deletion or archiving to inexpensive storage.
- Zombie VMs — Self-service provisioning is a great idea, until it isn’t. Most users assume there is no cost associated with spinning up a few VMs (and then forgetting about them), but zombie VMs consume resources needlessly.
VM sprawl leads to wasted storage and computing resources in the virtual environment. It’s inconspicuous at first, but it doesn’t remain that way for long and it eventually affects your performance and OpEx.
New guidebook: An Expert's Guide to Optimizing Virtualization Management
We’ll explore more concepts in optimizing virtualization management in the next two blog posts. Meanwhile, download our new guidebook, An Expert's Guide to Optimizing Virtualization Management. It highlights five important areas many companies overlook in their virtual landscape where, with a little digging, they can usually recover lost ROI and regain the promise of virtualization.