Data centers have undergone rapid change over the past few years. First there was the advent of virtualization, which allowed data centers to finally take advantage of their true capacity while reducing their footprint. Cloud solutions weren’t far behind as more and more business units found outside solutions that would actually meet a need better and cheaper than what could be done in-house.

All of this change has left many data centers with legacy equipment. While most of the time legacy servers get decommissioned, it’s not entirely uncommon for one to escape the watchful eye of IT personnel. It could be that a server was left up and running after a new application was implemented in order to support legacy data. It could be that a server was supposed to be decommissioned, but was left in service because it still played a secondary role (such as file services for a business unit).

Dead servers cost money

Whatever the reason, many data centers simply have a surplus of dead servers. These servers suck up resources such as power and cooling, and in some cases they can even soak up their share of administrative overhead.

It’s estimated that a server can cost an organization about $2,000 per year. Most experts suggest that somewhere between 5 and 30% of servers in a given data center are dead. For an enterprise with 1,000 servers, even the low end of that estimate means $100,000 a year in additional costs.

Getting legacy equipment offline will save the data center money, reduce power and cooling demands, and create a more efficient environment.

Grab control of your data center

There are several steps you need to take to identify and get rid of dead servers:

1. Create good documentation on every piece of physical equipment. That should include servers, switches, and more. Each piece should have ownership information as well as its data and network connections.
2. Don’t commission a new server until you know who’s responsible for it. Deployment needs to include this documentation.
3. Use this information to identify servers of an unknown status. Once you have servers identified, look at the processes and traffic on those servers. Contact potential stake holders, and ask about shutting it down.
4. If necessary, move ancillary services to another server. Chances are you can move file services somewhere else fairly easily, for example.
5. After the inquiry period, shut down dead servers. Keep them on site for a short time, allowing for the possibility that a stakeholder wasn’t properly identified.