The Insights module provides comprehensive monitoring for your Elastic Compute Service (ECS) instances, covering performance, resources, processes, and events.
Prerequisites
You have connected your application to ECS Insights.
Features
Host monitoring
Host monitoring provides a global overview of your ECS instances. You can switch to a single-instance view to view detailed metrics, including the following:
Infrastructure metrics
Process metrics
Advanced network metrics
GPU metrics
Process analysis
You can analyze processes from a global perspective across all your connected ECS instances. The analysis includes:
Cross-instance insights into the resource consumption of processes. This includes real-time comparison and time-series analysis of CPU, memory, disk, and network usage.
A detailed list of processes on an instance, including the process ID (PID), runtime, start command, and resource consumption.
Network insights for processes on an instance. This includes inbound and outbound network speeds, inbound and outbound packet counts, and the number of connections.
Statistics on the number and status distribution of child processes and subthreads for each process.
System calls made by processes, including I/O read and write speeds and the number of file descriptors.
Performance and capacity analysis
You can analyze resource consumption for compute, network, and load. You can select multiple instances to compare their resource usage. This analysis supports both real-time comparison and time-series modes.
Resource statistics
You can view a global statistical analysis of your ECS instance resources. The analysis covers the total number of instances, vCPUs, and memory in your account. For monitored instances, the analysis also includes the total number of instances, total vCPUs, total and remaining memory, and total and remaining disk space.
Analysis of monitored resources
Statistical analysis of the number, percentage, and resources of your monitored ECS instances.
Resource analysis by user
Statistics on the total number of ECS instances, requested vCPUs, and memory, aggregated by user.
Resource analysis by region
Statistics on the total number of ECS instances, requested vCPUs, and memory, aggregated by region.
Resource analysis by instance type
Statistical analysis of instance type distribution and a time-series analysis of the number of ECS instances for different instance types.
Comparative analysis of Linux instance resources
You can compare the remaining percentage of memory, CPU, and disk space across instances. This analysis is available in real-time and time-series modes.
Orange indicates that less than 50% of a resource remains. Red indicates that less than 30% of a resource remains.
Comparative analysis of Windows instance resources
You can compare the remaining percentage of memory, CPU, and disk space across instances. This analysis is available in real-time and time-series modes.
Orange indicates that less than 50% of a resource remains. Red indicates that less than 30% of a resource remains.
Event statistics
You can view a global statistical analysis of events for your ECS instances. This includes:
Statistics by event severity level, including CRITICAL, WARN, and INFO.
Statistics by event type, such as Cloud Assistant and Instance StateChange.
GPU card analysis
You can view global statistics and detailed analysis for your GPU-accelerated instances. This includes:
GPU card statistics: Shows the number of connected cards, a real-time comparison of metrics such as power and temperature, and a time-series trend analysis.
Resource details and trends: Shows detailed information for GPU-accelerated instances, including analysis of clock speed, power, temperature, and memory usage.