iot.eetimes.com, Apr. 03, 2019 –
In a recent article published to EETimes, author Rick Merritt wrote how Hewlett-Packard Enterprise (HPE) announced they are now using neural networks to better predict potential failures on the nearly four million hard disk drives in the company's InfoSight service monitors. This was no small undertaking, as Merritt outlines the amount of time and expertise required in order to get it up and running.
One of the first challenges the company faced with data curation on failures of select hard drive models. This task took between six and nine months to complete, during which engineers had to go through 2 petabytes of system data from the last two years. And this wasn't the company being overly detailed with their approach: if you don't have enough data, you cannot properly train a neural network.
Click here to read more...