In the previous post, we have looked at the significance of predictive technology. Let us now look at how it works for predicting drive failures. As with most predictive maintenance technologies, a lot of data is first collected representing healthy and failing drives under various operating conditions. This data is then used to develop algorithms that can help predict drive failures.

Life Prediction Scores

ULINK DA Drive Analyzer analyzes historical usage data from millions of drives to provide intelligent forecasts for potential drive failures. Based on machine learning algorithms, it comes up with a Life Prediction Score that indicates how healthy a drive is. The lower the score, the lower the health of your drive. For example, a Life Prediction score of 10 indicates a healthy and normal drive. When the Life Prediction Score drops below 9.5, the software would send out a warning through email notifications and alerts. A score below 4.5 indicates that the drive is in a critical state and should be replaced or backed up to avoid data loss.

Threshold-Based Alerts

However, predictive maintenance isn’t limited to machine learning. Simple threshold-based alerts can also predict whether failures are likely in the near future. Through constant monitoring of individual health metrics from S.M.A.R.T. and others, DA Drive Analyzer can send out threshold-based alerts that can serve as warnings of drive problems to come. Though threshold-based alerts may give more false positives than the machine learning algorithms that generate our main drive health predictions, they have the advantage of being able to pinpoint the root of problems with greater specificity.

Non-ML Features

ULINK DA Drive Analyzer also monitors non ML features such as S.M.A.R.T. attributes, temperature, and test IOPS. The analysis and insights from DA Drive analyzer are displayed on a diagnostics dashboard that provides deep insights at a quick glance. It can indicate how many drives are healthy, how many are at moderate risk, how many are at critical risk, and how many have health indicators that have crossed system-specific thresholds (faulty). Such information can help a user make an informed decision on when to replace fault-prone drives.

 

QNAP Launches the AI-Powered DA Drive Analyzer 2.0 – Predicts NAS Drive Failure Within 24 Hours & Enhances Enterprise Privacy

Photo Credit: Vladimir_Timofeev