Goodison: Zombies Are Killing Your Bottom Line
It’s been said that people are motivated by two primary emotions: greed and fear. These motivations certainly could hold true for C-suite executives when it comes to managing their data centers – and the “zombies” within them.
Featured in pop culture phenomena from the “Plants vs. Zombies” video game to the television series “The Walking Dead,” Zombies are everywhere today. Your data center, however, is the last place you want to see them.
The fact is that there's a good chance you are operating servers and other equipment that are, quite frankly, zombies. They’re achieving nothing as they sap energy, power and other resources from your business.
It’s estimated to cost at least $2,000 per year to support a single mid-tier server. Estimates also show that somewhere between 10 and 30 percent of most companies’ servers are “dead.” (Source: Federal Data Center Consolidation Initiative, Sept. 2011.)
This means that in a 4,000-server enterprise, for example, if even 400 of them are dead, you're looking at a bill of $800,000 annually – factoring in electricity, cooling and other operational expenses – for servers that are doing nothing. That’s more than a drop in the bucket to keep your zombies alive. But it’s an issue that is solvable.
What are the problems?
So what are the “greed” and “fear” triggers for executives in the data center space? On a financial level, it’s about cost, including those associated with everyday manpower, senior management time, space capacity, upgrades of licenses, energy and other capital and operating expense considerations. Reducing expenses associated with running unused data center servers will help increase efficiency, reduce the need to expand data centers – essentially doing more with the same – and maintain profit levels.
In addition to the financial considerations, CIOs – and their C-level colleagues – should be fearful of the potential security issues that zombie servers can create. Servers that are not being used are prime candidates to be taken over, because often they are unmonitored and unpatched.
How do your servers become zombies…?
Frequently, a server is commissioned by a business unit for a specific application or project. Eventually, the need for the server goes away, but a decommissioning process doesn’t take place because there is no tieback to the business unit using the server. Data center management is not certain which servers are unused and at risk of causing downtime, so the device is never fully decommissioned.
Furthermore, if the user of the service on the server was never recorded or the commissioning was done on an ad hoc basis, then no one knows who uses it. In time, the organization is aware there's a server physically there, but doesn’t know what it does or who provisioned it. Again fear of downtime (the unknown) means it is seen as less risk just to leave thing alone
…And how can you keep the zombies away?
So, how can you prevent the zombies from taking up residence in your data centers? For starters, it’s about good documentation. Keep an accurate record of your physical equipment, owner information, and network and data connections. Your record-keeping should even include switches, which can be zombies too. It’s important to note that even virtual machines can be zombies as well, this is not a problem just restricted to the physical world.
Also when rolling out a server, be clear on its purpose. Check how long the server is expected to be needed with the business unit. And from a processing point of view, look at what the server is doing. Are you seeing power changes, for example, or is it flat?
Not only should you establish good records at the outset, they must be updated regularly, as new equipment is added, including owner information. A good data center infrastructure management (DCIM) tool that provides a common, real-time monitoring and management platform will help considerably in this endeavor.
Part of the change process is having people document what they do in a more structured fashion. As such, you shouldn't put a new server into a live state unless you can identify who owns it. We see this as looking at deployment in a slightly more holistic way. Beyond physically deploying the server, you need to make sure you've got the documentation there as well.
Once the documentation and tools are in place, it’s important to regularly query and analyze the information at your disposal. This includes power draw, CPU utilization, network traffic and so on – as these metrics can be flat for dead servers.
As executives strive to do more with less and streamline operations, it's critical to have complete visibility into the data center and how it’s performing. Good management will pay dividends, not only in terms of ROI, but in alleviating the greed and fear associated with zombies.
Paul Goodison is Chairman & CEO of Cormant Inc., an information, communication and technology infrastructure management company engaged in the development and marketing of advanced infrastructure management software systems. He has more than two decades of experience managing major IT projects across the globe and has first-hand experience with the consequences of poorly managed infrastructure in large IT environments, which inspired him to co-found Cormant.