More moving parts mean more chance of failure. Replace “moving parts” by “comatose IT servers” and the adage still holds true. You may be tempted to reply that 1) there aren’t many of this kind of server anyway, and that 2) comatose servers may not be doing any good, but as such they are not doing any harm either. If so, get ready for a disaster recovery reality check on both counts!
Let’s take the first item. How many unused or comatose servers are there in the world? Between 8 and 10% is the conclusion of several studies. The Uptime Institute suggests 15 to 30%. One large data centre in the US even had a comatose server count of over half its installed base.
At the very least, these servers waste money because of maintenance contracts and software licensing costs that are money down the drain. That’s money that could be spent on improving disaster recovery for the active servers!
More perniciously, they put extra load on power and air conditioning plants. Not only is that distinctly un-green, it also increases the risk of plant overloading and failure, and therefore system outage (active as well as comatose).
And if nobody is bothering about them, comatose servers could make ideal staging posts for hackers to do data exfiltration, store malware in view of further attacks, and so on. So, for the second item, comatose servers represent a risk, even if they are not doing anything.
Decommissioning servers is therefore a wise and often necessary precaution. However, whether unused servers are destined for reuse through consolidation or definitive disposal, decommissioning has certain rules. It is possible that users or departments have stored data on comatose servers that they must recover.
These servers may also feature in link maps held by other servers, in which case their sudden disappearance might cause problems in the active servers. So, make your preparations as necessary, so that in trying to improve your overall disaster recovery, you don’t create more problems than you solve.