Monitoring & Alerting

Key Monitoring Dashboards

• Pipeline Health — success rate, average runtime, failure trends, SLA breach heatmap
• Cluster Utilization — CPU, memory, network I/O per cluster and per user
• SQL Performance — query latency P50/P90/P99, queue depth, data scanned per warehouse
• Cost Dashboard — DBU consumption by team, pipeline, user; month-to-date vs budget
• Data Quality — row counts, null rates, schema drift alerts across all pipeline outputs
• Security Events — login failures, permission denied events, policy violations

Creating Alerts

• 1. Navigate to Admin Console → Monitoring → Alerts → New Alert.
• 2. Select the alert type: Pipeline Failure, Cluster Cost, Query SLA, Data Quality, or Custom Metric.
• 3. Define the condition: metric, comparison operator, threshold, and evaluation window.
• 4. Configure notification channels: email, Slack, Microsoft Teams, PagerDuty, or webhook.
• 5. Set alert frequency: notify on every occurrence, or suppress for N minutes after first trigger.
• 6. Click Save Alert.

Alert Configuration Example

# Alert: Pipeline SLA breach
name: daily_sales_sla_breach
type: pipeline_duration
pipeline: daily_sales_pipeline
condition:
  metric: duration_minutes
  operator: greater_than
  threshold: 90  # Alert if pipeline takes more than 90 minutes
  window: last_run

notifications:
  - channel: slack
    target: "#data-ops"
    message: "⚠️ daily_sales_pipeline exceeded 90-minute SLA: {{ duration }} min"
  - channel: email
    to: [data-lead@company.com]
    subject: "[NATIS] Pipeline SLA Breach — daily_sales_pipeline"

cooldown_minutes: 60  # Don't re-alert for 60 min after first trigger