You can use Elastic Compute Service (ECS) monitoring features to identify and troubleshoot instance issues and address potential risks before they affect your business.
Handle system events in a timely manner
When the system performs O&M and identifies issues that affect the running of ECS instances, system event notifications are sent. System event notifications provide information such as solutions and event cycles. We recommend that you handle system events in a timely manner to prevent consequences of system events such as instance restart and stop from affecting your business deployed on the instances. For more information, see Overview.


Monitor the running metrics of instances
Alibaba Cloud collects and shows the running metrics of your instances to help you understand their real-time and historical running status. You can check whether instances are running normally based on their running metrics. If the CPU utilization of an instance is consistently high, you can check whether processes on the instance are abnormal or whether the configurations of the instance cannot meet your requirements.
- The following running metrics of an instance are displayed on the Instance Details page in the ECS console:
- The usage of computing, storage, and network resources such as the CPU utilization, disk read/write performance, and packet forwarding rate
- The CPU credit usage of a burstable instance
- The following running metrics of an instance are displayed on the Host Monitoring page in the CloudMonitor console:
- The usage of computing, storage, and network resources such as the CPU utilization, disk read/write performance, and packet forwarding rate
- The active processes on an instance
- The GPU memory usage of a GPU-accelerated instance
Use the alerting feature to trigger notifications
You can use the alerting feature of CloudMonitor to set alert rules for specified events and instance running metrics. When specified events occur or when instance running metrics are abnormal, notifications are sent to the contacts by email. This reduces manual O&M workloads. For more information, see Configure event notifications and Configure alerts for an ECS instance.

