Technology companies like Microsoft and Google are sprinting to build data centers and specialized processors for running services like cloud computing. But hidden behind the hardware is a network of components like surge protectors and batteries that feed and control power into corporate computers.
On Wednesday, Eaton, one of the largest makers of such power systems, released new software that helps to predict when components are about to overheat or fail. The software sends alerts to data center employees so that they can replace parts before the servers go down.
The software, PredictPulse Insight, collects data from the data center and sends it to the cloud, where algorithms scan for parts that might be about to falter. Every fifteen minutes, it reports temperature, load percentage, battery health, humidity, and other factors about uninterruptible power supplies.
If something appears wrong, Eaton will manually send alarms to an online dashboard and a mobile app for data center employees to review.
Using sensors and software programs to trawl for data center glitches is nothing new. Earlier this year, Cisco announced technology for tracking data moving inside computer servers so that technicians could quickly find malfunctions. Other big engineering companies like Siemens and Schneider Electric provide software to gauge power consumption and identify broken parts.
But Eaton claims that predicting component failure has been largely ignored. “We’ve applied big data innovation to remote power monitoring in a way that hasn’t been done with data center infrastructure,” said Del Misenheimer, senior vice president of Eaton’s Power Quality division, in a statement.