Identify and resolve unexpectedly high internal network traffic on Elastic Compute Service (ECS) instances monitored in Cloud Monitor.
Problem description
On the Host Monitoring page of the Cloud Monitor console, the inbound internal network traffic for an ECS instance is unexpectedly high, as shown in the following figure.
Cause
ECS instances use a public network interface controller (NIC) to provide services, so internal network usage is typically low. Internal traffic spikes when other ECS instances transmit large amounts of data to the instance.
Note Backend ECS instances of a Server Load Balancer (SLB) instance typically show high internal traffic because SLB communicates with backend instances over the internal network.
If no other ECS instances are transmitting data but internal traffic remains high, the instance may be infected with a virus that causes it to forward a large number of packets. For more information, see Solution.
Solution
Use the following methods to identify the processes consuming internal bandwidth on Linux and Windows instances.
- LinuxNote The NetHogs tool is an open source command line interface (CLI), which is similar to the top command in the Linux operating system. The NetHogs tool is used to analyze the real-time bandwidth usage of processes or applications.
- Download the NetHogs tool.
- Run the following command to install the NetHogs tool:
yum install nethogs
- Run the following command to view the internal bandwidth usage of each process:
nethogs eth0
- WindowsNote If the ECS instance runs Windows Server 2008 or later, you can view the internal bandwidth usage of processes in Resource Monitor.
- On the ECS instance, right-click the taskbar and select Task Manager.
- On the Process tab of the Task Manager dialog box, view the internal bandwidth usage of processes.