Several years ago when Cabling Installation & Maintenance began to sharply focus on the cabling and other systems within data centers, it did not take long before I noticed an overarching theme to just about every conversation I had on the topic. It didn’t matter what we began talking about—for example, overhead versus underfloor cable routing, the copper/fiber mix in data center networks, managing huge bundles of cables at the rack—the conversation always ended up coming around to managing the facility’s heat and keeping critical equipment cool. The theme made its way into virtually every conversation.
Recently the theme of an overheated data center was much-discussed at the offices of Wikipedia, to be certain. In late March, an overheating problem at the online encyclopedia’s European data center caused a temporary global shutdown of the site. The Wikimedia Technical Blog explained that servers in the European data center shut themselves down in order to avoid damage from overheating. Once that happened, the site moved all user traffic to its Florida data center; however it then was discovered the failover mechanism in the Florida location that changed Domain Name System entries, was broken.
The problem was resolved quickly, but many users could not access the site for more than an hour. The blog made no mention of what efforts the organization makes to control the heat in its European data center. But something tells me a man by the name of Ian Seaton would have a field day in that data center. (What? You don’t know who Ian Seaton is? Go to our Web site, www.cablinginstall.com, and search his name.)
The biggest snafu in the Wikipedia situation was that the site’s backup plan did not work successfully. It sounds as if things got resolved pretty quickly, and no one was inconvenienced for a terribly long time, but the overheated data center would have been seamless to users if the backup systems had worked as they were expected to. It begs the question: What is your backup plan? And how sure are you that it will work when called upon? It doesn’t matter if you’re a data center manager, network administrator, or the operator of a contracting organization. Someday, sometime, the systems you rely upon for business operation will fail. Do you know what you’ll do when that happens?
If you don’t, 90 minutes without Wikipedia will be a best-case scenario. Think: Travelers stranded in Europe and at the mercy of Eyjafjallajokull.