New Autodetect, from Splunk Application Performance Monitoring (APM), uses machine learning to significantly reduce manual effort and improve accuracy for your service alerts. Autodetect establishes performance baselines for every service, creates automatic detectors based on sudden changes in latency, errors, and request rates, and allows engineers to customize and subscribe to notifications for alerts on these detectors. As a result, engineers reduce time and effort in configuring their alerts, and receive the most accurate alerting across their services.
Less Manual Effort and Smarter Alerting
Historically, creating, updating, and modifying service alerts has been a manual process that can lead to toil and inefficiency, especially as the amount of services grows in a distributed system. For example, engineers using static thresholds to establish normal error rates might be flooded with alerts as their system scales. Also, as the number of services increases, especially in microservices and cloud-native environments, configuring and updating individual alerts becomes significantly more time intensive.
Splunk Autodetect for APM is different. It uses machine learning-based algorithms to analyze service performance, establishing baselines, thresholds, and patterns for every service. Next, we provide automatic detectors for sudden changes in latency, errors, and request rates (aka RED metrics). Engineers can simply subscribe to recommended alerts, customize trigger thresholds, create filters, and control who receives notifications.
Autodetect in Action:
To use APM AutoDetect, you must first send traces from your application to APM. Once you start sending your traces to APM, AutoDetect alerts and detectors automatically appear on both the APM UI and the Alerts & Detectors page.
To learn more about using and customizing autodetect alerts, see documentation.
Try it Today
Autodetect helps you monitor service performance as you deploy faster and troubleshoot problems across cloud native environments. See the documentation to get started.
— Mat Ball, Director of Product Marketing, Observability
