Microsoft explains Azure's hardware failure prediction system and its notifications

Microsoft says Azure Failure Prediction and Detection (AFPD) uses machine learning and telemetry to predict, mitigate, and remediate hardware failures before they interrupt customer workloads.

In an Azure blog post, Microsoft says AFPD has cut hardware-related VM interruptions by 36% and raised the monthly node repair rate by 80%.

AFPD runs automatically across the Azure fleet and is not a paid tier or something customers configure at the infrastructure layer, according to Microsoft. Customers can receive notifications about hardware events through Resource Health, Resource Graph, Event Grid, and Scheduled Events. Live migration is available on many, but not all, M-series VMs, and Microsoft directs customers to its feature support table to check.