Here are a few we see more often than we’d like:

☛ A rack of servers loses power when an IT administrator unintentionally overloads an already maxed-out power strip.

☛ A large data center virtualizes and consolidates its most critical applications on a cluster of servers. Using the failover mechanism of the virtualization platform, they feel protected from hardware failure. Unfortunately, in their panning, they don’t recognize that each of the servers is dependent on the same UPS, which means that if the UPS fails, no UPS-protected servers are available to migrate the affected loads to.

☛ An operator is trying to determine whether power capacity that was just exceeded on a rack is only an anomaly or a developing trend. She goes on “gut feel” and leaves it alone. The next time power capacity in that rack is exceeded, a breaker trips and all the servers downstream of that breaker that are running mission critical applications are suddenly shut down.

☛ In a large. mission critical data center, the provisioning and installation of servers is so complex that only highly paid contract engineers are able to perform the task.