How do people in your organization regard virtual hardware? Perhaps before we answer that, we should back track for a moment and ask a different question: how do people in your organization regard physical hardware? That’s probably an easier question. I know that when I think of “physical” hardware the word “asset” comes to mind. And with assets comes justification, procurement, paperwork, and an ongoing maintenance responsibility. In short… it’s expensive and there is usually plenty of process to keep sprawl in check… at least somewhat.
Now let’s go back to the original question about how people regard “Virtual” hardware. Well… they are virtual right? What does virtual mean? In most people’s mind… you could associate the word with the idea of “not real.” Virtual Reality, for example, is not reality. So, Virtual Machines are not machines? Right? And, if they aren’t machines, they aren’t assets… which means they grow on trees?
Ok, I’m probably rambling a bit here, but you get the point. In my organization, we use VMWare ESX in our test lab for software QA and development. The original purpose of bringing virtualization into the lab was to collapse and consolidate costly, aging, heat-bearing assets, into virtual entities that could be more easily managed with a smaller footprint. This isn’t a unique goal for most organization adopting virtualization. So, what happened? We currently have all of the assets that we had previously, plus two ESX servers, plus more assets.
Why did this occur? I don’t know, we are still trying to figure that out… but consider the following to the extent that it is applicable in your organization:
- Virtual machines are easy and fast to deploy. We have templates define that make it even easier in a QA/DEV environment to setup and tear down machines on the fly. Something happened… people started getting what they needed to run their testing when they asked for it. When people get what they want… they want more. You all read that book about the mouse, the cookie, and the glass of milk, right?
- Because machines are “virtual” the marginal cost of adding another one is minimal, at least in the short term.
So how do we “manage” this problem. My suggestion is to virtualize your assets. Apply the same scrutiny for virtual hardware as you do for physical hardware. Here are some tips that have helped us get back on track and get the most out of our virtual resources:
- Manage sprawl of VM’s the same way that you would manage sprawl of anything else. If resources are needed temporarily, insist on a timeframe commitment (”I only need it for a little while…” is not good enough).
- Be specific with machine requirements. Virtual resources can be easily squandered away.
- Have a process in place to reclaim/re-allocate/destroy expired virtual assets.
- Don’t over-extend; keep some resources in buffer for “emergencies.”
