CloudWatch Alarms

• Trigger notifications for any metrics
• Various options (sampling, max, min)
• Alarm States
  • OK
  • INSUFFICIENT_DATA
  • ALARM
• Action
  • Stop Terminate, Reboot or recover EC2 instance
    • For recovery we do a satus check
    • We can do recovery for same private, public, ip, metadata or placement group
  • Trigger Auto Scaling Action
  • Send notification to SNS (which you can pretty much do anything)
• High resolution alarm: 10 seconds or 30 seconds trigger
• Regular laram: any multiple of 60 seconds trigger
• You can test the alarm using set-alarm-state CLI command