The metrics of an event-triggered task for system monitoring are measured at the scaling group level. The metric value is the average value of monitoring data for all instances in the scaling group. When the number of instances in a scaling group changes, the metric value is also updated. This topic describes the metrics that are supported by event-triggered tasks for system monitoring.

Monitoring metrics of ECS instances

The following table describes the metrics of ECS instances that are supported by event-triggered tasks for system monitoring.

Collection source Metric Unit Applicable network
ECS instance CPU Utilization % Virtual Private Cloud (VPC) and classic network
Outbound Internal Network Traffic KB/min VPC and classic network
Inbound Internal Network Traffic KB/min VPC and classic network
Outbound Public Network Traffic KB/min VPC and classic network
Inbound Public Network Traffic KB/min VPC and classic network
System Disk Read (Byte/s) None VPC and classic network
System Disk Write (Byte/s) None VPC and classic network
System Disk Read (IOPS) None VPC and classic network
System Disk Write (IOPS) None VPC and classic network
CloudMonitor agent CPU Utilization % VPC and classic network
GPU Utilization % VPC
Idle GPU Memory Percentage % VPC
GPU Memory Utilization % VPC
Memory % VPC and classic network
Average System Load None VPC and classic network
Total TCP Connections None VPC and classic network
Established TCP Connections None VPC and classic network
Internal Network Packets Sent None VPC and classic network
Internal Network Packets Received None VPC and classic network
Public Network Packets Sent None Classic network
Public Network Packets Received None Classic network

If you select CloudMonitor as the collection source, Auto Scaling installs the CloudMonitor agent on all instances in the associated scaling group. You can enable New Purchase ECS Automatically Installs CloudMonitor in the CloudMonitor console to ensure that the monitoring data of newly purchased ECS instances can be collected.

If you want to select GPU metrics, make sure that GPU drivers are installed on GPU-accelerated instances.

Metrics of elastic container instances

The following table describes the metrics of elastic container instances that are supported by event-triggered tasks for system monitoring.

Collection source Metric Unit Applicable network
CloudMonitor agent CPU Utilization % VPC
Memory Usage % VPC