How does Azure detect potential failures early and give customers options for responding to those signals before mission-critical workloads are affected?
Blog authored by: Sai Vaidhyanathan, Ayberk Ozturk, Ronit Sharma, Goran Condric, Holger Bruchelt, Andrew Boyd II
In this SAP on Azure video podcast, Microsoft engineering experts take a closer look at Azure Failure Prediction and Detection (AFPD), exploring how machine learning, fleet-wide telemetry, and proactive mitigation help reduce hardware-related interruptions. The discussion walks through AFPD's end-to-end approach to predicting, detecting, mitigating, and remediating hardware failures, including how customers can monitor infrastructure health and respond to notifications.
Watch the video below to learn how Azure is advancing platform reliability, with SAP workloads providing a practical example of these capabilities in action. Goran Condric and Holger Bruchelt host as Ayberk Ozturk discusses the motivation and vision for AFPD, Andrew Boyd gives an overview of the key services and how they work together, and Ronit Sharma presents a demo of how to proactively consume and react to these notifications
Learn more
- Previous AFPD blog: Revolutionizing Reliability: Introducing the Azure Failure Prediction and Detection (AFPD) system
- Research paper: AFPD: System for Prediction, Mitigation and Remediation of Node Interruptions in Cloud Infrastructure
- Resource Health virtual machine Health Annotations - Azure Service Health | Microsoft Learn
- Consuming notifications: Resource Health virtual machine Health Annotations - Azure Service Health | Microsoft Learn
Microsoft