Published on · Updated on
Corrected EC2 metric intervals and exceptions, separated custom metric resolution, and updated cost, alarm, scaling, and setup guidance.
For an Amazon EC2 instance, basic monitoring sends most instance metrics to CloudWatch every five minutes; detailed monitoring sends supported instance metrics every minute for an additional charge. The choice controls how often EC2 publishes its metrics. It does not add operating-system memory or free-disk metrics, and it is separate from the standard or high resolution you can choose when publishing custom metrics.
AWS documentation calls the EC2 options basic and detailed monitoring. This guide compares those EC2 settings, explains exceptions, and shows when a one-minute signal is useful. Other AWS services can use different metric intervals and separate options with names such as enhanced monitoring.
CloudWatch receives metrics from AWS services and from applications or agents you configure. Each metric has a namespace, name, dimensions, and data points with timestamps. The service or publisher determines which data it sends; an alarm then evaluates those data points over a period you choose.
For EC2, the basic-versus-detailed setting controls the reporting frequency of supported metrics in the AWS/EC2 namespace. It is not a universal switch for every metric in your account. AWS describes its service-specific options in the CloudWatch guide to basic and detailed monitoring.
Basic monitoring is the default for EC2 instances and has no CloudWatch charge for the EC2 metrics it provides. Most supported instance metrics, such as CPU utilization and network bytes, arrive in five-minute periods. EC2 disk read and write metrics describe instance store volumes; EBS storage has separate metrics.
| Metric source | Cadence on EC2 basic monitoring | What to remember |
|---|---|---|
| Most supported EC2 instance metrics | 5 minutes | Includes CPU utilization and many network and instance-store metrics. |
| EC2 status checks | 1 minute | Status check metrics are available at one-minute frequency without enabling detailed monitoring. |
| CPU credit metrics | 5 minutes | Credit metrics for burstable instances remain five-minute metrics, even when detailed monitoring is enabled. |
| Attached EBS volume metrics | Separate cadence | The instance's detailed-monitoring setting does not change EBS volume monitoring. Attached EBS volumes publish their own metrics. |
These exceptions matter when choosing an alarm period. A one-minute status-check alarm can work on a basic-monitoring instance, but lowering a CPU alarm's period does not create one-minute CPU data. Check the EC2 metric reference for the cadence and availability of each metric; some metrics are fixed at five minutes or are available only under particular conditions.
For EBS, distinguish volume metrics in the AWS/EBS namespace from the EC2 instance-level EBS metrics. AWS says attached EBS volumes automatically send one-minute volume metrics, and enabling EC2 detailed monitoring does not affect attached-volume monitoring. See the EBS metrics reference.
Detailed monitoring changes supported EC2 instance metrics from five-minute to one-minute reporting. It can also provide additional EC2 dimensions, such as aggregating metrics by instance type or image, for comparisons across groups of similar instances. It does not turn on guest operating-system metrics.
CloudWatch does not receive EC2 memory utilization, filesystem free space, or per-process measurements from the basic or detailed EC2 setting. To collect those, install and configure the CloudWatch agent. Agent metrics are separate from EC2 detailed monitoring and may incur custom-metric and API charges. For a practical EC2 agent setup, see our EC2 CloudWatch monitoring guide.
EC2 basic and detailed monitoring describe how frequently EC2 publishes its service metrics. Standard resolution and high resolution describe CloudWatch metrics you publish, such as application metrics. They are related to cadence, but they are different settings.
| Setting | What it controls | Typical granularity and use |
|---|---|---|
| EC2 basic monitoring | Cadence of supported EC2 instance metrics | Most metrics: 5 minutes |
| EC2 detailed monitoring | Cadence of supported EC2 instance metrics | Most supported metrics: 1 minute |
| Standard-resolution custom metric | Resolution selected when publishing a custom metric | 1-minute granularity; the publisher still controls when it sends data |
| High-resolution custom metric | Resolution selected when publishing a custom metric | Stored with 1-second resolution; alarms can evaluate at 10, 20, or 30 seconds |
High resolution does not make EC2's built-in metrics arrive every second. It applies to metrics published as high resolution, for example with the CloudWatch API's storage-resolution setting. You need to publish useful data at the cadence your decision requires; storing a metric at one-second resolution does not make a slower publisher send more often. AWS explains metric resolution and retention in the CloudWatch concepts guide; the PutMetricAlarm API lists supported 10-, 20-, and 30-second high-resolution alarm periods. For publishing custom metrics, see our CloudWatch custom metrics guide.
Use basic monitoring when five-minute trends are enough for capacity review, troubleshooting, or alarms that do not need sub-five-minute EC2 data. It is also a reasonable default for instances whose important failure signal is a one-minute EC2 status check.
Enable detailed monitoring when a supported EC2 metric needs one-minute data, such as a CPU-based alarm or a scaling decision that should respond sooner to load changes. A shorter interval can reduce the wait for the next datapoint; it does not guarantee an alert or scaling action within one minute. Alarm evaluation, missing data, instance warmup, launch time, and the application itself also affect response time. For threshold and evaluation choices, see our CloudWatch alarm guide.
For EC2 Auto Scaling, AWS recommends detailed instance monitoring for scaling policies that use EC2 instance metrics, because one-minute instance data supports a faster response to changing load. Auto Scaling group metrics are a separate option: enable group metric collection when you need one-minute group-level capacity metrics. Review the Auto Scaling instance monitoring guidance and our EC2 Auto Scaling setup guide.
If the decision needs sub-minute application data, EC2 detailed monitoring is not enough. Publish an appropriate high-resolution custom metric and choose a 10-, 20-, or 30-second alarm period when that evaluation cadence is useful. If the metric is already emitted by another AWS service, check that service's documentation instead of assuming the EC2 setting changes it.
Basic EC2 monitoring and EC2 status-check metrics have no additional CloudWatch metric charge. EC2 detailed monitoring is charged according to the number of metrics the instance sends and the time they are sent. Metric counts vary by instance type; charges are prorated, and pricing can vary by Region and usage tier. AWS currently lists 10 custom and detailed-monitoring metrics per month in the CloudWatch free tier, shared across those metric types rather than granted per EC2 instance. Check the current offer terms for your account.
Do not estimate detailed monitoring as a fixed price per instance or apply one universal custom-metric rate. Check the current CloudWatch pricing page for your Region and account, then use the AWS Pricing Calculator for an estimate. Alarms, agent or application metrics, API requests, and logs may add separate charges. EC2 detailed monitoring itself does not add a PutMetricData request charge.
Detailed monitoring changes how often supported EC2 data points arrive; it does not extend CloudWatch's metric history. CloudWatch retains one-minute data for 15 days, then aggregates it to five-minute resolution through day 63 and hourly resolution through day 455. High-resolution custom data points with periods under one minute are available for three hours before being aggregated. Choose the collection cadence for decisions you need to make now, and account for aggregation when you need older history. See AWS's metric retention details.
You can manage detailed monitoring for an existing running or stopped instance in the EC2 console: select the instance, choose Actions → Monitor and troubleshoot → Manage detailed monitoring, then enable or clear the option. The change is per instance. For an Auto Scaling group, also set the monitoring option in its launch template so new instances use the intended setting.
The AWS CLI commands for an existing instance are:
aws ec2 monitor-instances --instance-ids i-0123456789abcdef0
aws ec2 unmonitor-instances --instance-ids i-0123456789abcdef0
Replace the example ID with the instance ID. Enabling and disabling use the EC2 MonitorInstances and UnmonitorInstances permissions. At launch, the EC2 CLI accepts --monitoring Enabled=true on run-instances. See AWS's EC2 detailed-monitoring instructions.
Do not assume the EC2 basic/detailed switch applies to RDS, S3, or other services. For example, Amazon RDS sends its standard database metrics to CloudWatch at one-minute periods by default. RDS Enhanced Monitoring is a separate option that sends operating-system metrics to CloudWatch Logs at a configurable interval. Each service has its own metric set, cadence, options, and pricing.
When monitoring a mixed architecture, check the metric reference for each service and confirm the metric's namespace and period before creating alarms. The CloudWatch service can store metrics from different sources together, but it does not make their publishing behavior identical.
No. Install the CloudWatch agent to collect operating-system memory, filesystem, and process metrics. EC2 detailed monitoring only changes the reporting cadence of supported EC2 metrics.
Yes, for EC2 status-check metrics, which are available every minute on basic monitoring. For most other basic EC2 metrics, use a period of at least five minutes. A one-minute alarm period cannot create one-minute source data; use detailed monitoring when a supported EC2 metric needs one-minute reporting.
No. EC2 detailed monitoring changes supported EC2 service metrics to one-minute periods. High-resolution custom metrics are published with one-second resolution and can support 10-, 20-, or 30-second alarm periods. They are separate settings with separate costs.