Running production EC2 instances limited to 5-minute basic monitoring

Five-minute metrics hide short spikes

The EC2 detailed monitoring guide sets out the difference. Basic monitoring, the default, publishes status checks every minute and all other metrics in 5-minute periods at no charge. Detailed monitoring publishes metrics in 1-minute periods and is charged per metric. On a production instance, a CPU spike that lasts two minutes can vanish inside a 5-minute average, and alarms or scaling policies keyed to those metrics react up to five minutes late.

Finding running instances on basic monitoring

aws ec2 describe-instances \
  --filters Name=instance-state-name,Values=running Name=monitoring-state,Values=disabled \
  --query 'Reservations[].Instances[].[InstanceId,InstanceType,Tags[?Key==`Name`]|[0].Value]' \
  --output table

The gates before a finding

Three things must be true. The instance must be running. ZopNight must have read its monitoring setting from AWS and seen that detailed monitoring is off; if the setting was not reported, nothing fires. And the instance must not look like a development or test machine.

How non-production instances are excluded

Because this rule asks you to spend money, ZopNight skips instances whose name or environment tag (such as env, environment, stage or tier) marks them as dev or test. An explicit production environment tag overrides that, so a box named test-runner but tagged environment=prod is still evaluated.

A cost increase, not a saving

This finding never adds to savings totals. Enabling detailed monitoring raises spend: CloudWatch pricing charges detailed monitoring metrics as custom metrics, $0.30 per metric per month for the first 10,000, prorated by the hour. The pricing page’s own example assumes 7 metrics per instance, which works out to $2.10 a month; the real count depends on instance type.

monthly cost of enabling = metrics sent by the instance x $0.30

Turning on detailed monitoring

1. Select the instance in the EC2 console.
2. Choose Actions, Monitor and troubleshoot, Manage detailed monitoring, and enable it. From the CLI: aws ec2 monitor-instances --instance-ids i-0123456789abcdef0.
3. Update alarms and scaling policies to use 60-second periods so they benefit from the finer data.