Proactively monitor health of your business using Amazon Lookout for Metrics
summary
Amazon releases Amazon Lookout for Metrics in preview: a fully managed ML service that ingests time series from S3 intake (also CloudWatch, Redshift, Salesforce, Marketo), runs hourly or backtest analyses, and surfaces anomalies with impact analysis and root‑cause hints. Models learn patterns and adapt (on‑the‑go or from historical data); alerts include severity score 87 and a default sensitivity threshold of 70 adjustable to 65–85. Outputs push via SNS to PagerDuty and Lambda, Slack, Datadog or SMS. NextRoll beta tests it on pixel logs (near‑billion daily impressions), finds both a mobile timeout bug in UK/Germany and an unexpected Android traffic spike.
AI-written summary · about
chapters (14)
- 0:00Introduction to monitoring
- 1:08What is anomaly detection
- 2:38Pain points with traditional tools
- 5:12Amazon’s anomaly detection journey
- 8:48Introducing Lookout for Metrics
- 9:33How Lookout for Metrics works
- 10:39Benefits of the service
- 13:29Use cases and demo setup
- 14:36Demo creating detector
- 17:30Demo results and impact analysis
- 19:40Tuning and alert setup
- 21:32NextRoll’s challenges and solution
- 25:24NextRoll’s experience with the service
- 27:09Next steps and resources
related talks
38mAnomaly detection in the cloud: how to detect issues in data [...]
25mTransform your business with data-driven insights
01h 01mBetter reliability and uptime with analytics and observability
14mChaos Genius: Open Sourcing Outlier Detection and Root Cause Analysis
35mObservability’s Sixth Sense: Detecting Anomalies in Metrics
26mServerless Architecture for Product Defect Detection
31mHow AIOps Drives Operational Resilience on AWS
09mUsing your technical skills with Amazon Ads