Cloud Fleet Health
Fleet Overview
The Cloud Fleet Health dashboard provides a real-time view of infrastructure health across all connected cloud accounts. Resources are organized by account, region, and service type. Health indicators use a traffic-light system (green/yellow/red) based on configurable threshold rules applied to CloudWatch, Azure Monitor, and GCP Cloud Monitoring metrics.
- CloudWatch: Cross-account CloudWatch metrics sharing enabled from connected accounts to account
373544523367 - Lambda:
StackFlowFleetHealthCollectoron 5-minute EventBridge schedule - DynamoDB:
StackFlow_FleetHealthSnapshottable for historical health state - SNS:
stackflow-fleet-alertstopic for health degradation notifications
The fleet health view is the first place on-call engineers check during a P1 incident investigation. It shows which resources are currently degraded, what metrics triggered the degradation status, and links directly to the relevant cloud console for immediate investigation.
Health Metrics
| Resource Type | Key Metrics | Default Thresholds |
|---|---|---|
| EC2 / VM | CPU utilization, memory, disk I/O | Warning: 80% / Critical: 95% |
| RDS / Aurora | CPU, connections, storage, replication lag | Warning: 75% CPU / Critical: 90% |
| Lambda | Error rate, throttles, duration, concurrent executions | Warning: 1% errors / Critical: 5% |
| ElastiCache | CPU, memory, connections, evictions | Warning: 70% memory / Critical: 85% |
| API Gateway | 4XX/5XX rate, latency, throttle count | Warning: 1% 5XX / Critical: 5% |
Alerting Integration
Fleet health alerts integrate with CloudWatch Alarms (AWS), Azure Monitor Alerts, and GCP Cloud Monitoring alerts. When any of these fire, StackFlow receives the alert via the configured webhook or SNS subscription and creates (or updates) an incident in the ITSM module automatically. The alert payload is stored in the incident description and used by the AI triage engine for initial classification.
Drill-Down View
Click any resource in the Fleet Health view to open the resource detail panel. The detail panel shows: current metric values, 24-hour metric trend charts, open incidents for this CI, recent changes, and Neptune graph of dependent services. The panel also provides quick actions: create incident, start remediation workflow, or open in cloud console.
Historical Trends
Historical fleet health data is stored in S3 as Parquet files and queryable via Athena for long-term trend analysis. The Fleet Health dashboard shows 7-day and 30-day trend sparklines for each resource. Full historical data export is available in the Report Studio for custom analysis periods.